Visual Analytics and Data Mining Techniques

This assignment delves into the realm of visual analytics and its integration with data mining techniques within Human-Computer Interaction (HCI). It examines various approaches to visual analytics, including information visualization and progressive visual analytics, emphasizing their role in exploring and understanding complex datasets. The assignment also explores data mining techniques used in HCI, such as clustering and classification, highlighting their applications for tasks like user modeling and behavior analysis. Finally, it discusses the challenges and opportunities presented by these methods in the context of modern HCI systems.

DATA ANALYSIS
Name:
Student ID:
Supervised by:
Date:

Abstract
Music has significant role in all the events in the world. The current research revolves around the
Argentine Tango dance party or events and the DJs’ from all the corners of the world. The main
objective of this research is to measure the factors which holds key role in selection of music
track’s list for worldwide DJs, to play in the Tango events, and to certify these music tracks with
high influence, from the Data Analysis scope. Thus, this research aims to answer- “What can
data analysis reveal from the selection of music tracks by DJs in Tango music events and how?”
which helps in enhancing the event. This research supports to answer the factors which influence
the track selection and in which degree. Data analysis tool is used for clarifying the patterns. TF-
IDF (term frequency–inverse document frequency), data mining and data visualisation methods
are used in this research. SPSS software is used for data analysis of the current research. The
current research has another set of objectives i.e., to observe the correlation between the
provided music data and then to qualify the type of music played for entertaining the crowd in
various geographical regions, and for different events. The overall goal of the research method is
to answer the following questions- How does the preferences change over time? Find the region
which will probably begin with a new trend? Determine the duration for the average lifecycle of
a hit i.e., from peak to decay? Identify the highly played orchestras in all the continents? How
will the dance tempo impact the event from beginning to end? Recognize the most often or
highly played music tracks in all the continents and how are these tracks compared? Does there
exists any pattern for music track selection based on the event type or day? Mention the
resemblance or differences of track selection, from different DJs of different parts of the world?
Initially, the research gathers the track playlists from different DJs present in various parts of the
world. Next step includes data processing with the help of data analysis i.e., data mining through
visualization. Visualization is this research’s product. This research provides music data
visualization. The beneficiaries of this research are determined. The respective results are
represented and discussed briefly. The limitations of the current research are highlighted.
Further, the reflections of the research, its methods and procedures are reflected in the later part
of the research. The achievement and future work of the research are also presented.

Table of Contents
1. Introduction............................................................................................................................1
1.1 Background......................................................................................................................1
1.1.1 Why this research?..................................................................................................3
1.1.2 Information Visualization.......................................................................................4
1.2 Research Questions.........................................................................................................4
1.3 Why did I choose the Project.........................................................................................4
1.4 Beneficiaries.....................................................................................................................5
1.5 Products of this Research...............................................................................................5
1.6 Research Objectives........................................................................................................5
1.7 Scope and Definition.......................................................................................................6
1.8 Methods............................................................................................................................7
1.8.1 Software Development Methodology.....................................................................7
1.9 Testing the Objectives.....................................................................................................8
1.10 Structure of Dissertation................................................................................................8
2. Critical Context......................................................................................................................9
2.1 Approach..........................................................................................................................9
2.2 Visual Analytics.............................................................................................................10
Process of Visual Data Analytics............................................................................................11
Building blocks of Visual Analytics........................................................................................11
2.3 Visual Data Mining.......................................................................................................12
3. Method...................................................................................................................................13
3.1 Participants....................................................................................................................13
3.2 Materials........................................................................................................................14
3.3 Methods..........................................................................................................................14

3.4 Procedures.....................................................................................................................14
4. Results....................................................................................................................................18
5. Discussion.................................................................................................................................38
1.1 Validity of Results.........................................................................................................41
1.2 Generalization of Results..............................................................................................42
2. Evaluation, Reflections and Conclusion.............................................................................43
2.1 Literature Review.........................................................................................................43
2.2 Reflection on Topic and Objectives.............................................................................43
2.3 Reflection on the Selected Methods.............................................................................44
2.4 Reflection on the Plan...................................................................................................44
2.5 Accomplishment............................................................................................................45
2.6 Conclusion......................................................................................................................46
2.7 Future work...................................................................................................................47
References.....................................................................................................................................47

1. Introduction
This project is the result of data collection and visualization, which aims to deliver
trusted and quality information. All the people in the world attract to good music, and the
involvement of audience in the events can change the reputation of the conducted event. The
current research project concentrates on generating music data visualization, for helping the DJs
in the music track’s selection process, by analyzing the effectiveness of the selected music track
playlist for the Tango music events.
This research includes convincing the DJs to contribute help in the selection of music
tracks, for the Tango music events. Next, it requires gathering of data, data mining and reporting
the results. Once the visualization is provided successfully, it could contribute to enhancing the
tango music event. This research aims to measure the factors which plays a prominent role to
select the music track-list, and to certify these music tracks such that it increases its influence. It
also foresees to observe the correlation between the supplied music data and to qualify the type
of music played for meeting the crowd’s needs. Thus, the music industry uses data mining
method for enhancing the experiences of their audience. This research opts SPSS software for
completing the data analysis.
1.1 Background
This section highlights the basic knowledge of the current research, which assists in
following the conducted research. Music is a vital part of all the events, worldwide. This research
revolves around the Argentine Tango dance party, also known as “Milonga”. The DJs have a
major role in this party, as they play four selected tango music tracks. But, at times, they just
play three tango tracks. A set of songs is referred as “Tanda”. The DJs play short snippets of
non-tango songs, amid tandas which are known as “Cortinas”. In general, the Tango festivals are
annual events, which are conducted for many days around the weekend. Popular dancers and
instructors are part of this event. In general, the tandas consists of various rules and regulations,
which completely depends on the DJs. The DJs has all the rights to play the tango playlist as per
their wish. Therefore, here data mining is required to scrutinize the playlist, which is collected
from various DJs of worldwide. The collected music data is decreased for declining the risk of
threat. The collected music track playlist will contain specific information, to ensure data mining
with the help of visualization could generate optimal results. For the data analysis, orchestra and
1

the track name has significant importance. Every single playlist should contain its location, time,
season and the event type they were played for. This type of details will help in retrieving
accurate results. Furthermore, the information like, audience type and DJ’s background acts as
complementary information which helps in reducing the risks of errors in the results. Various
tango events can have positive or negative impact due to the DJ’s music selection. It is stated that
the local milongas could be held periodically or regularly or it can be conducted as a special
event.
The Tango festivals comprises a mix of instructional classes and workshops, it also
includes the instructors’ performances and milongas, which grasps the attention of the dancers
who come from all the parts of the world. These dancers are well experienced. In such social
events, the DJs are referred as key entertainer, and makes the crowd enjoy their played music
tracks. The Tango marathons has similar schedule like the other type of events, where the only
exception is that they omit classes or workshops and it mainly attracts seasoned dancing. The
DJs regulate the event with their selected music track, which maintains crowd’s good mood and
they consider it as their responsibility to keep the whole event enjoyable. This observed
requirements of this research is to perform data mining, to take right decision by right insight for
completing the research process. To help this research, SPSS software is utilized for
accomplishing data analysis.
Some of the playlists that are utilized are as follows:
1) Budapest marathon is dancer level 3 (1-5) in Budapest.
2) Avrig can be listed as casual festival with dancer level 3, in Romania.
3) Cuib is a local Milonga, dance level 2 in Romania
4) Sibiu is festival, dancer level 3, in Romania
5) Anivertango is marathon, dancer level 4, Romania
The DJ belongs from Romania, who is a female aged 30 years.
The locations selected for this research are listed below:
1) London
2) Miami US
3) Providence US
2

4) Brussels
5) Florida Keys US
The provided even type is mentioned in the below section:
a) Local Milonga
b) Local practica
c) Evening millonga
d) Outdoors versus. not outdoors
Experience dancers (Five) and unexperienced dancers (one) are considered, as follows:
Providence US: 2
Miami US: 3
Florida Keys: 4
Brussels: 3
London: 3
The expectations of this research includes- The DJs requires generating a track playlist
with factors that influence the Tango events, to select right music track. The playlist must depend
on the stereotype, for various factors and must correspond with the analysis’s inputs. The factors
could be the even type, geographic location of the event and crowd. It is also noticed that it can
be checked for the experience of the event and how it varies over time.
1.1.1 Why this research?
This research is used for providing a famous data mining method for identifying the
similarities, patterns and correlation. The earlier tools failed to provide effective results. But, the
currently utilized data mining method for visualization is, comparatively instant and more
effective. In case of accomplishing this research, it furnishes visualization for enhancing the
capacity to explore the data. Thus, this research will then generate appropriate music playlist’s
stereotypes, which is based on multiple factors and corresponds with the factors like event type,
audience and geographic location, type of event, audience and its variation with time. However,
this outcome influences the Tango event’s DJs. On the other hand, it helps the music industry,
3

which enhances experiences of the audience and in turn increases their profit. For added profit, it
is required that the music track must match the seasonal trend and place. And this requires
correlation, which can be found by data mining process. Further, it handles large data collection
and also identifies various patterns from it.
1.1.2 Information Visualization
Information Visualization is also known as InfoVis. It represents a type of study that has
visual representations of abstract data which could be numerical and non-numerical. For
instance, text or geographic information. Thus, this type of data helps to have interactivity, it also
provides clear understanding and it can identify the difficult data patterns. InfoVis is an online
source which is reachable to large number of research audience (Sindiy et al., 2013). It
effectively and instantly identifies the anomalies along with the data patterns. It is a tool
especially used for large dataset (Robertson and Kaptein, 2016).
1.2 Research Questions
The important question of this research is to resolve: “What can data analysis reveal from
the selection of music tracks by DJs in Tango music events and how?” Hence, this question’s
answer determines whether, the data analysis of the provided playlists being played in Tango
events- from DJs around the world, could help in enhancing it, then in tango event which factors
could influence the track selection and in which degree. The patterns are clarified based on data
analysis tool.
1.3 Why did I choose the Project
For visualization, music data is an interesting platform. My personal opinion on
visualization can help to improvise the data for conveying and to present to the analysts for
easing their work. On the other hand, it has the flexibility to visually recognize and manipulate
the data, by addition and removal of new insights on the data. Tango events are very popular and
it includes contribution of various DJs from different parts of the world, for gathering the music
track playlist/ data. In contains various events and with different crowd. Satisfaction of the
crowd, by entertaining them is the ultimate goal of this event. If the crowd is not satisfied then it
4

can decline the popularity of the tango events. Thus, music data visualization is a challenging
task for the researcher.
1.4 Beneficiaries
This research includes the following groups of people as beneficiaries of this study:
1) The Tango dancers are the first group, who will find the information of this
research more informative.
2) The next group includes, the worldwide DJs for the Tango events. As, this
research considers the DJs’ major task i.e., music track list of the event for
ensuring the crowd to retain their dancing mood. On the other hand, this research
can also help them to play these tracks in various events, in different parts of the
world.
3) The final group refers to the sociologists, who will be informed about how the
DJs in different occasion attempt to assume the crowd in the events. The obtained
results will interest to know about various inspiring factors for the DJs’ to select
the music by considering the sociological patterns such as, continent, type of
event and dancing level. Collaboration with sociologist and the data analysis
results can conclude the sociological considerations.
Eventually, this research should provide music data visualization.
1.5 Products of this Research
The current research’s product is a visualization, which projects visual data mining
methods for enabling selection of music tracks, based on certain sociological considerations.
1.6 Research Objectives
This research’s key objective is to measure the factors that has a major role in selecting
the music track list, and to certify these music tracks in a way that it has higher influence, from
the scope of Data Analysis. It also foresees to identify the correlation between the supplied music
data; then find answers for when, where, to whom and by who; and to qualify the type of music
played for entertaining the crowd based on different sociological factors like different crowd,
different place and different event.
5

The related requirements to meet the research objective (RO) are listed below:
RO1: How does the preferences change over time?
RO2: Find the region which will probably begin with a new trend?
RO3: Determine the duration for the average lifecycle of a hit i.e., from peak to
decay?
RO4: Identify the highly played orchestras in all the continents?
RO5: How will the dance tempo impact the event from beginning to end?
RO6: Recognize the most often or highly played music tracks in all the continents
and how are these tracks compared?
RO7: Does there exists any pattern for music track selection based on the event
type or day?
RO8: Mention the resemblance or differences of track selection, from different
DJs of different parts of the world?
The other objectives are:
i. Collect enough track playlists from different DJs present in various parts of
the world.
ii. Perform data processing using data analysis (i.e., data mining through
visualization), which answers the research question.
iii. Determine the research beneficiaries.
iv. Identify the limitations of the research.
v. Design visualization solution.
vi. Implement visualization solution.
vii. Evaluate the results.
viii. Provide reflections of the conducted research.
1.7 Scope and Definition
This study concentrates on visualization, like how one or more visualizations could be
improved and support the DJs around the world, for the tango events to select the effective track
playlist. The prototype will concentrate on selecting the track playlist, and to certify these tracks
in with high influence, using Data Analysis. This research considers visualization based on
different sociological factors like different crowd, different place and different event, to find the
6

correlation between the provided music data and to qualify the music to be played for satisfying
the crowd.
1.8 Methods
The current research requires problem-driven research approach, because it helps in
resolving the real-world issues. The music track selection is also a real-time event which requires
lot of working, as it could be challenging to meet certain circumstances like the geographical
area, different crowd and different event. The event must have appropriate music, which matches
needs of the place, crowd and event.
To perform data analysis, SPSS software is utilized in the current research. For data or
playlist processing, data mining method is used for gathering data. Because, it has the capacity to
handle handles large data and also recognizes various patterns from the data collection. This
research requires finding the correlation among the gathered data, which is easy with data
mining. The correlation attempts to find the match with seasonal trend and location of the event,
for enhancing the experiences of the audience.
Once the data is collected, data cleaning takes place. As the received playlists from the
DJs of different parts of the world will not have same required format. Thus, data cleaning
amends the format of the gathered data, which recovers the missing elements. The gathered data
is visualized by data visualization. Then data analysis is performed with Data Visualisation, to
increase the capacity to recognize the influencing factors, for music track selection. The method
like TF-IDF (term frequency–inverse document frequency) efficiently supports information
retrieval. It helps the process like text mining that acts as a weighting factor. This research uses
text mining as the music track playlists are in the format of text file. The research answers are
revealed by combining the elements and its frequencies.
1.8.1 Software Development Methodology
Any field requires effective method, which changes from time to time. Similarly,
appropriate selection of music track is also important for the music event organizers.
Development requires implementing an effective methodology. This research’s objective is to
develop music data visualization prototype. Later, using this prototype, a software will be
developed to help the beneficiaries. The prototypes are developed in small divisions, where
every single division is an iteration or iterative prototyping that contains a new set of prototypes.
7

All these prototypes are first developed and approved by the professionals. Various prototypes
are initially designed by the developers, for refining the best prototype. However, it is time
consuming and difficult, but it is the best way to meet the desired results.
1.9 Testing the Objectives
For ensuring the right answer for the research question, it is essential to meet all the
objectives of the research. The answer can be retrieved by the researcher, just by having a clear
picture of the issues related to data exploration, for identifying the requirements. The identified
requirements are utilized for designing and implementing the prototype, with the help of data
mining methodology. For ensuring to test the research question, it is necessary that the designed
prototype must meet the identified requirements and implement the paradigm of data
exploration. Finally, the prototype is utilized for answering the research question. The objective
of the test is to evaluate the prototype’s capabilities for visual data mining, when compared to the
present screener which hardly has any capabilities to perform visual data mining.
1.10 Structure of Dissertation
Chapter 1 is the introduction of the dissertation about the research. This chapter provides
the background of the overall research, where it describes the reason for establishing this
research and its objective, where a list of research objectives for the research are listed. This
section also sheds light on the scope, methods and beneficiaries of the research.
Chapter 2 includes critical context, which is an overview of the approaches, visual
analytics and visual data mining.
Chapter 3 revolves around the methods used in the research. This chapter helps in clearly
understanding the methods used, research procedure, and the participants of the research. The
stepwise methods implemented in the research are studied. The importance of the implemented
methods are determined in this section.
Chapter 4 projects the results of the research. The research questions are answered in this
section. The execution results are represented clearly. This chapter is a significant part of the
dissertation, which displays the results of the conducted research using the SPSS software.
Chapter 5 ensures discussion of all the acquired results. The determined results are
discussed briefly in this part of the dissertation. All the research questions are answered, based
8

on the analyst’s point of view. The results are determined based on SPSS software and data
mining method.
Chapter 6 is the final part of this dissertation, which includes Reflection on the selected
method, Reflection on the Plan. Later, the research accomplishments are listed and the research
is concluded in the conclusion part. The necessary recommendation for the encountered
limitations are scrutinized and presented in the recommendation area. Finally, the future work is
forecasted for the current research.
2. Critical Context
2.1 Approach
The Approaches that can be used for data analysis can be divided into 2 different
categories as Quantitative Approach and Qualitative Approach of Analysis. The approaches that
are considered to be Quantitative interprets the results of the statistics in the terms of numerical
values in accordance with the certain degree of confidence level. The confidence level can also
be changed by us manually. Quantitative analysis approaches can help in collecting the
qualitative information in an organized way. In this project, the analysis has to be performed on
the playlists that are given to the event. The approaches that can be performed for analyzing the
data is the co-relation analysis (Qualitativedataanalysis.net, 2017). The co-relation analysis is
very important in this analysis which can find the relation or similarity between the playlists. By
considering the similarities only, the playlists can be separated into different categories. Good
planning of performing the analysis on data can help in minimizing the energy and time
consumption. In this project, we adopt a qualitative data analysis which is the approach of
considering the raw data for analysis and making explanations on those data for research. The
objective of this approach is to analyses the data and for examining the meaningful content that
is present in the data. The relation between the data is found using this approach. The type of
Qualitative data that is taken for the analysis is the Music Playlist. The different types of analysis
that can be performed in the qualitative data analysis are Content Analysis, Narrative analysis,
Discourse analysis, framework analysis and the Grounded theory analysis (Health.herts.ac.uk,
2017). The different types of approaches that can be used in the process of data mining are
Association, Classification, Clustering, prediction, Decision Trees and Sequential pattern. So out
9

of these different approaches of the data mining techniques, the approach that is suitable for the
analysis of this project has to be taken (ZenTut, 2017).
2.2 Visual Analytics
Visual Analytics is the science of analytical reasoning that can be supported by the visual
interfaces. The data that is produced today is increasing day-by-day and the capability of
collecting and storing the data has also been increased in an incredible rate (Stolper, Perer and
Gotz, 2014). The collection process becomes trivial in field of the data mining. The collected
data may be incomplete or unstructured. To make the incomplete or unstructured data to a
complete or structured form, different kinds of processes are taken place. Data Cleaning, Data
Transforming and data preprocessing are carried out to make the data to be structural
(SearchBusinessAnalytics, 2017). The methods that are performed in the visual analytics permits
the persons who makes decisions for combining the flexibility, creativity and the knowledge of
the human with more amount of storage capacity. It is used for enhancing the capabilities of the
computers by gaining the insight into the difficult problems. The visual interfaces can be used for
making interactions between the person and the computer. Visual Analytics is considered as an
approach that can be helpful in combining the visualization, data analysis and human factors.
The below represented triangle specifies the areas of research that are related to Visual
Analytics. Out of these three areas of the research, the human factors play a significant role in
making interactions between the human and the computer. The visual analytics makes profits in
the data analysis fields by using the methodologies called Retrieval of information, management
of data and the representation of knowledge. The visual analytics can also be defined as group
that can be used for measuring the system and process which combines the information
visualization and the analytical reasoning (Techopedia.com, 2017). It is based on the premise in
which the visualization process is used as a mediator for analyzing the data. The system of visual
analytics will make use of the software dashboard for presenting the results of analytics visually.
The interface of visual analytics makes the users to understand the results in an easy way
(Vismaster.eu, 2017). Visual Analytics is very important in the areas where analysis has to be
performed for huge amount of data. The main fields where the visual data can be applied are
physics, climate monitoring, biology, security and commerce (Blascheck et al., 2016).
10

Process of Visual Data Analytics
The Visual Analytics is the process of combining the automatic and visual analysis
methods by interacting with humans for gaining the knowledge from the data. The first step in
the process of visual analytics is preprocessing and transforming the data for deriving the
different representations of the data. The tasks that are involved in the preprocessing of data are
Data cleaning, Data normalizing, grouping the data and the integration of the heterogeneous
sources of data. After preprocessing the data, the analysis methods has to be applied. The
analysis methods can be either visual or automatic. The method of analysis depends on the type
of data that is going to be analysed. If the analysis method called automated analysis is used in
the beginning then the methods of data mining are applied for generating the models of the actual
data. Visualization process permits the analyst to make interactions with the methods of analysis
that are automatic by changing the parameters or choosing different analysis algorithms. Model
Visualization can be used in evaluation of the findings of the models that are generated.
Building blocks of Visual Analytics
1) Visual Analytics is an approach which is used for integrating science and technology
from different disciplines.
2) Visualization is defined as the heart and it is the major component of the analysis.
3) This process not only helps in communicating with the data values but also for
monitoring the processes like data management and data mining (Information
visualization and visual analytics: challenges and opportunities, 2013).
4) Infrastructure and Evaluation are significant things in all disciplines. Infrastructure can be
used to link all the processes, services and functions that are in need for the applications
of the visual analytics. Evaluation is the process of assessing the efficiency and user
11

acceptance of the methods and models in a defined and organized way (Ruf and Schwab,
2016).
2.3 Visual Data Mining
Visual Data mining is defined as the process of interaction and the analytical reasoning
with the representations of the abstract data visually. The process leads to the process of
discovery of robust patterns visually in those data or to provide some more guidance for applying
various data mining methods and techniques of analytics. The process of visual data mining
depends on the interconnectivity of the tasks, selecting the visual representation, set of
manipulations that are interactive in nature, the techniques that are involved in the analytics. It is
a new kind of approach that can be used for exploring the large datasets by combining the
methods that are traditional and visualization techniques of information. Visual data mining can
be broadly classified into 3 different categories as follows (Yu, Yurovsky and Xu, 2011).
1) Visualization of Data Mining Process
2) Visualization results of Data Mining.
3) Interactive Visual Data mining.
Visual Data Mining Methodology
Visual Data Mining process involves 3 step process. The first step involved in that process
is the Overview, second thing is zoom and filter, the third one involved in the 3 step process is
the details on demand. The data analyst will collect the overview of data. In this overview, the
interesting patterns or groups of the data are identified. The techniques of visualization are very
useful in showing the overview of data. Visual data mining is also known as hypothesis process
of generation. The data visualization process allows the analyst for gaining the insight into the
12

data. The visual data mining is used to allow data exploration in faster rate. The visual data
exploration techniques are used for providing a higher amount of customer satisfaction and the
confidence level which are known as the findings of the exploration. This paves the way for the
high demand for the techniques of visual exploration. The Process of Visual data mining depends
on the automatic part, algorithm called data mining and the technique of visualization which is
known as the integral active part. The common approaches that can be used for realizing the
various visual data mining approaches. The different approaches of human involvement in the
field of visual data mining are preceding visualization, Subsequent Visualization and tightly
integrated visualization (Keim, 2002).
3. Method
3.1 Participants
DJs are the main participants of this research. DJ (Disc Jockey) is a person who mix the
recorded songs from different sources. The selection of the track is mainly based on the decision
making process of the DJ. This decision making process will be based on the results of the
research questions. There are different types of Disk jockey like Radio DJs who introduce the
play music in FM, AM, internet radio or digital stations. Club DJs play the music in
discotheques, nightclubs, parties, events in stadiums and raves. Mobile DJs plays the recorded
music in a variety of events from portable sound systems. The DJs are categorized by their
musical performance techniques. This technique is depend on their situation. Now a Days the
music technology is improved and different types of equipment’s are used by the DJ's to mix and
play the music. Traditionally, DJ mixer is used to mix the vinyl records. Then compact discs
became a famous media to publish the music with high quality. But now most of the CDJs use
the SD cards or USB flash drives to play the digital music. Most of the people uses laptop,
tablets and smartphone computers. Hence the DJs began to use the software’s with specialized
controller hardware and sound cards. Different types of techniques are used by DJs for the better
mix of blend music. This technique also includes auto mixing, equalization and other sound
sources. The frequency and complexity of the technique is mainly based on the settings in which
the DJ is going to work. Club DJs focus on the advanced techniques than radio DJ's. They mostly
rely on the smooth transition of the song by using different techniques.
13

3.2 Materials
The materials which are used in this data analysis are based on the set of songs which are
played in the tango events. This set of song is collected from the famous DJ's around the world.
These songs are played in different events like local evening outdoors, Tango marathon 2015,
local outdoor afternoon 2017, local evening 2013 and US practica evening 2011. It also contains
different type of play lists like Budapest marathon, avrig and cuib. This song list contains
different variables such as Artist, Album name, Title, length, year, genre, Bit rate and media. The
data analysis is mainly based on the six variables. They are Type of event, Country or continent
of the event, Days of a marathon, Time of milongas, Origin of the DJ and Audience or dancers of
the events.
The data’s are collected in multiple formats and they are converted to the universal format
by using the data cleaning method. The successful flow of the data analysis is based on the track
of play list from different DJ's. Different assertions and clarifications are made with the DJ's.
The DJ's play list is copyrighted with its legal author.
3.3 Methods
The method which is used in this research is listed below.
Data mining
1) Data cleaning
2) Data Visualization
3.4 Procedures
The work breakdown structure of this research is given below.
14

Convincing DJs to Contribute
DJs are the main participants of this research. This research is based on the data analysis
of different music tracks which are played in the tango events. Hence the first and foremost step
of this research is to get the play lists from the DJs. So we have to convince them to contribute
their play list for our research. It is not possible to continue our research without play lists. The
tango events held in different locations and regions around the world. To get those play lists
which are played in those events, we have to convince the DJs of those events to proceed our
research.
Data Gathering
Data gathering or Data collection is the next step of our research to get the play lists from
the DJs of the tango events. In this step the play lists are collected from the DJs who are
accepted to contribute their play lists for our research. The data’s collected from different
locations and regions such as Brussels, Florida, London, Miami and US. These play lists contains
different albums, titles, Bitrate and their ratings.
Data Mining
Data mining process is used to extract the hidden information’s from the large data bases.
It is one of the most powerful technique which has the high potential to focus on the most
important data in the data warehouse. It also predicts the behaviors, trends, knowledge-driven
15
Research
1.
Convincing
DJs to
Contribute
2. Data
gathering
4. Report3. Data
mining
Data
cleaning
Data
Visualization

decisions of the data. It also answers the business questions which requires more time to resolve.
It focus on the hidden patterns of the database and finds the predictive information.
This data mining technique is rapidly implemented in the existing hardware and software
platforms to improve the value of the existing information. It is also integrated with the new
products and systems. It analyses the massive techniques and delivers the answers for the
research and business questions within a short amount of time. This data mining technique is
mostly used in product development and long process of the research. This evolutionary process
begins when the dataset is stored in the computer and requires to improvements to access the
data set. This datamining technique is supported by three technologies such as massive collection
of data, data mining algorithms and powerful multiprocessor computers.
The main scope of this data mining process is to search the valuable information from the
large database. It also generates the new technologies and provides opportunities with following
capabilities.
Automatic prediction of behavior and trends
This data mining process automates the predictive information finding process in a huge
database.
Automatic Discovery of unknown patterns
This data mining process sweep the data base to find the hidden patterns in the previous
step. It also detects the problems in the pattern recovery process and identifies the anomalous
data in the dataset.
This data mining technique also gives more benefits through the automation process in
the existing hardware and software platforms and it can be implemented in the new systems to
upgrade the development process. If this data mining tools are implemented in the parallel
processing systems with high performance, it will analyze and find the results within minutes.
This faster processing gives more experiments for the user to understand about the complex
dataset. High speed data processing gives practical for the user to analyze the huge dataset.
In our research, the dataset collected from different tango events are needs to be
preprocessed to improve the quality of the dataset. Data cleaning and Data visualization are the
two sub process of data mining.
16

Data Cleaning
Data cleaning is the very first step in data mining. It is used to understand, correct and
qualify the data to get accurate result in the data analysis process. It is mainly used to discover
the incomplete and inaccurate data based on the business analysis. If the data is not qualified it
will lead to an inaccuracy and wrong conclusion. Hence we are using data cleaning process to
get the accurate result from the data analysis.
The given play list in not in the required format. Hence we have to do data cleaning process to
change the play list in a perfect format. This method also recovers the missing elements,
inconsistencies and outliers from the play list.
Data Visualization
Data visualization is the next step in the data mining process which is used to present the
information in a Visual format. This format gives the clarity for the further understanding of the
information. It can be represented in the form of 2 dimensional graph, 3 dimensional topology
based on the country.
It is a paramount and efficient method to display the information in a graphical format.
This method is used to get the correct result from the data analysis. Frequency of appearance is
the main variable in this data analysis which displays the playlist in all tracks and it is displayed
as a number. This number has different variables such as event type, country or continent of the
audience, DJ, milongas time and marathon day. The analysis of this information will be used to
improve the influence factors of the music.
Term frequency-inverse document frequency
It is one of the popular method which is mainly used for the information retrieval in data
mining process. This method is used as a weighting factor in text mining process. This method is
also used in this research and it holds the main resource of this project such as play list of the
music track which is in the CSV format. It is the optimal method for clustering the play list.
Term frequency and inverse document frequency is used to measure the importance of the
word in the document and it gives the collection of document. If the word is frequently appeared
in the document then it should be an important word and we should give high priority for the
word. But if that word is appeared in many documents then it is not a unique identifier. Hence
we have to give low score for the word. The frequency of the elements are combined with the
other elements weight to form a complete cluster to reveal the answer for the research questions
17

like most played orchestra, songs, selection of music in different continents, song preferences of
the DJ's.
4. Results
London_Local_Outdoor_Afternoon_2017
The dataset collected from the London local outdoor afternoon _2017 event is analyzed in the
SPSS software to find the results of the research questions. Initially the data set is analyzed to
find the preferences of the song and how it change over the time. This data set contains the
details of song such as artist, title, album, Year, Genre, Rating, Bitrate, Path, length and Media.
Two variables such as Ratings of the song and Album names are used to find the preferences of
the song. C graph analysis method is used to find the results. The result shows the simple bar
graph with Albums and their mean rating. The resultant graph shows that 16 albums have 100
ratings and the others albums are below 100. From this data set we can find that these 16 albums
have high preference than the other.
Rating vs. Album
C Graph analysis
18

Average (Mean)
To find the average life cycle of the hit such as from peak to decay, Descriptive analysis method
is used. Rating is taken as a variable. The descriptive statistics table shows the result of the
average life cycle of the hit. The descriptive analysis shows the average life cycle of the hit with
mean value 52.68.
Descriptive Analysis
Descriptive Statistics
N Minimum Maximum Mean Std. Deviation
Rating 80 -1 100 52.68 49.088
Valid N (listwise) 80
Most played album
To find the most played orchestras in each continent, Pie chart is used. The resultant chart shows
the Album name and how many times they are played in this London local event. The values are
denoted in different colors. From this bar graph we can find that the most played album is
coleccion 78 RPM1940-42.
19

Tempo (Bitrate)
To find the tempo of the dance from starting to ending of the event, two variables such as Album
name and Bitrate is used. Simple line graph is used to show the results which shows the tempo of
the event from starting to ending of the event based on the bitrate of the album. From this line
graph we can notice the tempo of the dance from starting to ending. The tempo changes
frequently based on the album. 1940-1950 album has high bit rate.
Tracks
To find the most played track in each continent, pie graph is used. It shows the Song title based
on the number of times which are played in the event. From this pie chart we can notice the most
played track is Pocas Palabras.
20

Correlation between Music data and its quality
The correlation between the music data and the qualities are analyzed and the correlation table is
given below. Pearson correlation is used with four variables of the music such as year, rating,
bitrate and media.
Correlations
Year Rating Bitrate Media
Year Pearson Correlation 1 .485** .488** .b
Sig. (2-tailed) .000 .000 .
N 80 80 80 0
Rating Pearson Correlation .485** 1 .376** .b
Sig. (2-tailed) .000 .001 .
N 80 80 80 0
Bitrate Pearson Correlation .488** .376** 1 .b
Sig. (2-tailed) .000 .001 .
N 80 80 80 0
Media Pearson Correlation .b .b .b .b
Sig. (2-tailed) . . .
N 0 0 0 0
**. Correlation is significant at the 0.01 level (2-tailed).
b. Cannot be computed because at least one of the variables is constant.
21

Miami_2013_Evening_Local
The dataset collected from the Miami_2013_Evening_Local event is analyzed in the SPSS
software to find the results of the research questions. Initially the data set is analyzed to find the
preferences of the song and how it change over the time. This data set contains the details of
song such as artist, title, album, Year, Genre, Rating, Bitrate, Path, length and Media. Two
variables such as Ratings of the song and Album names are used to find the preferences of the
song. C graph analysis method is used to find the results. The result shows the simple bar graph
with Albums and their mean rating. In this dataset 30 albums have 100 ratings and other 7
albums have ratings below 100. From this dataset result we can notice that 30 albums which
have high preference.
Rating vs. Album
C Graph analysis
22

Average (Mean)
Descriptive analysis
To find the average life cycle of the hit such as from peak to decay, Descriptive analysis method
is used. Rating is taken as a variable. The descriptive statistics table shows the result of the
average life cycle of the hit. The resultant descriptive statistics table shows the average value of
the life cycle as 67.78.
Rating
Descriptive Statistics
N Minimum Maximum Mean Std. Deviation
Rating 94 -1 100 67.78 46.307
Valid N (listwise) 94
Most played album
To find the most played orchestras in each continent, Pie chart is used. The resultant chart shows
the Album name and how many times they are played in this London local event. The values are
denoted in different colors. From this pie chart we can notice the played album is Grandes Del
Tango 27.
23

Bit rate
To find the tempo of the dance from starting to ending of the event, two variables such as Album
name and Bitrate is used. Simple line graph is used to show the results which shows the tempo of
the event from starting to ending of the event based on the bitrate of the album. The resultant line
graph shows the albums with different bit rates and how the tempo changes from starting to
ending. Malena album has high tempo than the other albums.
24

Tracks
To find the most played track in each continent, pie graph is used. It shows the Song title based
on the number of times which are played in the event. From this pic chart we can find that the
most played track is Un Placer.
25

Correlation between Music data and its quality
The correlation between the music data and the qualities are analyzed and the correlation table is
given below. Pearson correlation is used with four variables of the music such as year, rating,
bitrate and media.
Correlations
Length Year Rating Bitrate Media
Length Pearson Correlation 1 .493** .702** -.334** .b
Sig. (2-tailed) .000 .000 .001 .
N 94 94 94 90 0
Year Pearson Correlation .493** 1 .526** .104 .b
Sig. (2-tailed) .000 .000 .331 .
N 94 94 94 90 0
Rating Pearson Correlation .702** .526** 1 -.085 .b
Sig. (2-tailed) .000 .000 .427 .
N 94 94 94 90 0
Bitrate Pearson Correlation -.334** .104 -.085 1 .b
Sig. (2-tailed) .001 .331 .427 .
N 90 90 90 90 0
26

Media Pearson Correlation .b .b .b .b .b
Sig. (2-tailed) . . . .
N 0 0 0 0 0
**. Correlation is significant at the 0.01 level (2-tailed).
b. Cannot be computed because at least one of the variables is constant.
ProvidenceUS_Practica_2011_Evening
The dataset collected from the ProvidenceUS_Practica_2011_Evening event is analyzed
in the SPSS software to find the results of the research questions. Initially the data set is
analyzed to find the preferences of the song and how it change over the time. This data set
contains the details of song such as artist, title, album, Year, Genre, Rating, Bitrate, Path, length
and Media. Two variables such as Ratings of the song and Album names are used to find the
preferences of the song. C graph analysis method is used to find the results. The result shows the
simple bar graph with Albums and their mean rating. In the resultant graph we can find that 23
albums have 100 ratings and the other albums have below 100 ratings. Hence we can conclude
that these 23 albums have high preference than the other albums.
Rating vs. Album
C Graph analysis
27

Average (Mean)
To find the average life cycle of the hit such as from peak to decay, Descriptive analysis method
is used. Rating is taken as a variable. The descriptive statistics table shows the result of the
average life cycle of the hit. From this descriptive statistics table we can find that the average
value of the hits is 82.75.
Descriptive Statistics
N Minimum Maximum Mean Std. Deviation
Rating 72 -1 100 82.75 37.770
Valid N (listwise) 72
Most played album
To find the most played orchestras in each continent, Pie chart is used. The resultant chart shows
the Album name and how many times they are played in this London local event. The values are
denoted in different colors. From this pie chart we can find that the most played album is Title
(Troilo).
Bitrate
To find the tempo of the dance from starting to ending of the event, two variables such as Album
name and Bitrate is used. Simple line graph is used to show the results which shows the tempo of
28

the event from starting to ending of the event based on the bitrate of the album. From this Line
graph we can notice the Robin’s Canaro album has high tempo than the other albums.
Tracks
To find the most played track in each continent, pie graph is used. It shows the Song title based
on the number of times which are played in the event. From this pie chart we can find that the
most played track in this region is Le Bien Pager.
29

Correlation between Music data and its quality
The correlation between the music data and the qualities are analyzed and the correlation table is
given below. Pearson correlation is used with four variables of the music such as year, rating,
bitrate and media.
Correlations
Length Year Rating Bitrate Media
Length Pearson Correlation 1 .485** .394** -.606** .b
Sig. (2-tailed) .000 .001 .000 .
N 72 72 72 63 0
Year Pearson Correlation .485** 1 .052 -.266* .b
Sig. (2-tailed) .000 .662 .035 .
N 72 72 72 63 0
Rating Pearson Correlation .394** .052 1 .152 .b
Sig. (2-tailed) .001 .662 .235 .
N 72 72 72 63 0
Bitrate Pearson Correlation -.606** -.266* .152 1 .b
Sig. (2-tailed) .000 .035 .235 .
N 63 63 63 63 0
Media Pearson Correlation .b .b .b .b .b
Sig. (2-tailed) . . . .
N 0 0 0 0 0
**. Correlation is significant at the 0.01 level (2-tailed).
*. Correlation is significant at the 0.05 level (2-tailed).
b. Cannot be computed because at least one of the variables is constant.
4. Brussels_2016_local_evening_outdoors
The dataset collected from the Brussels_2016_local_evening_outdoors event is analyzed
in the SPSS software to find the results of the research questions. Initially the data set is
analyzed to find the preferences of the song and how it change over the time. This data set
contains the details of song such as artist, title, album, Year, Genre, Rating, Bitrate, Path, length
and Media. Two variables such as Ratings of the song and Album names are used to find the
preferences of the song. C graph analysis method is used to find the results. The result shows the
simple bar graph with Albums and their mean rating.
30

Rating vs. Album
C Graph analysis
Average (Mean)
To find the average life cycle of the hit such as from peak to decay, Descriptive analysis method
is used. Rating is taken as a variable. The descriptive statistics table shows the result of the
average life cycle of the hit. From this Descriptive statistics table we can find that the average of
the hits is 92.29.
Descriptive Statistics
N Minimum Maximum Mean Std. Deviation
Rating 94 -1 100 92.29 23.190
Valid N (listwise) 94
Most played album
To find the most played orchestras in each continent, Pie chart is used. The resultant chart shows
the Album name and how many times they are played in this London local event. The values are
denoted in different colors. From this pie chart we can find that the most played album is
coleccion 78 RPM 1940-42.
31

Bit Rate
To find the tempo of the dance from starting to ending of the event, two variables such as Album
name and Bitrate is used. Simple line graph is used to show the results which shows the tempo of
the event from starting to ending of the event based on the bitrate of the album. The resultant
Line graph shows the tempo of the dance from the starting to ending with different peak values.
In which the instrumentals inolvidables has the highest bit rate (500) and it has high tempo.
32

Track
To find the most played track in each continent, pie graph is used. It shows the Song title based
on the number of times which are played in the event. From this resultant pie chart we can
conclude that the most played track of this dataset is Instrumentales Inolvidables.
33

Correlation between Music data and its quality
The correlation between the music data and the qualities are analyzed and the correlation table is
given below. Pearson correlation is used with four variables of the music such as year, rating,
bitrate and media.
Correlations
Year Bitrate Media Rating
Year Pearson Correlation 1 -.164 .a .037
Sig. (2-tailed) .189 . .764
N 68 66 0 67
Bitrate Pearson Correlation -.164 1 .a -.090
Sig. (2-tailed) .189 . .413
N 66 105 0 85
Media Pearson Correlation .a .a .a .a
Sig. (2-tailed) . . .
N 0 0 0 0
Rating Pearson Correlation .037 -.090 .a 1
Sig. (2-tailed) .764 .413 .
N 67 85 0 94
a. Cannot be computed because at least one of the variables is constant.
5. FloridaKeysTangoMarathon_2015_friday_afternoon
The dataset collected from the FloridaKeysTangoMarathon_2015_friday_afternoon event is
analyzed in the SPSS software to find the results of the research questions. Initially the data set
is analyzed to find the preferences of the song and how it change over the time. This data set
contains the details of song such as artist, title, album, Year, Genre, Rating, Bitrate, Path, length
and Media. Two variables such as Ratings of the song and Album names are used to find the
preferences of the song. C graph analysis method is used to find the results. The result shows the
simple bar graph with Albums and their mean rating. From this resultant graph we can find that
42 albums have 100 ratings and the others 10 albums have below 100. These 42 albums have
high preference than the other albums. Los Provincianos-1931/1934 Trio Ciriaco Ortiz-
1931/1955 album has more preference than others.
34

Rating vs. Album
C Graph analysis
Average (Mean)
To find the average life cycle of the hit such as from peak to decay, Descriptive analysis method
is used. Rating is taken as a variable. The descriptive statistics table shows the result of the
average life cycle of the hit. The resultant Descriptive statistics table shows that the average
value of the hits is 98.18.
Descriptive Statistics
N Minimum Maximum Mean Std. Deviation
Rating 99 70 100 98.18 5.222
Valid N (listwise) 99
Most played album
To find the most played orchestras in each continent, Pie chart is used. The resultant chart shows
the Album name and how many times they are played in this London local event. The values are
denoted in different colors. From this below chart we can find that the most played album is Los
35

Provincianos-1931/1934 Trio Ciriaco Ortiz-1931/1955.
Bit Rate
To find the tempo of the dance from starting to ending of the event, two variables such as Album
name and Bitrate is used. Simple line graph is used to show the results which shows the tempo of
the event from starting to ending of the event based on the bitrate of the album. From this line
graph we can find that Los Provincianos-1931/1934 has bit rate (441) than the other albums.
36

Tracks
To find the most played track in each continent, pie graph is used. It shows the Song title based
on the number of times which are played in the event. From this pie chart we can find that the
most played track of this dataset is Un Placer.
37

Correlation between Music data and its quality
The correlation between the music data and the qualities are analyzed and the correlation table is
given below. Pearson correlation is used with four variables of the music such as year, rating,
bitrate and media.
Correlations
Year Bitrate Media Rating
Year Pearson Correlation 1 -.213 .a -.154
Sig. (2-tailed) .077 . .215
N 71 70 0 67
Bitrate Pearson Correlation -.213 1 .a .129
Sig. (2-tailed) .077 . .216
N 70 114 0 94
Media Pearson Correlation .a .a .a .a
Sig. (2-tailed) . . .
N 0 0 0 0
Rating Pearson Correlation -.154 .129 .a 1
Sig. (2-tailed) .215 .216 .
N 67 94 0 99
a. Cannot be computed because at least one of the variables is constant.
5. Discussion
This chapter presents the primary activities of the current research, which are compared
with the objectives of the research. The comparison takes place to evaluate whether the
objectives were achieved or not.
Requirements Gathering and Validation
Initially, the necessary requirements for the research were gathered from the DJs of
various parts of the world. Next, the gathered data (playlist) was validated by using different
methods. Data mining process is used to validate the collected dataset and the inaccuracies,
insufficient data, impurities of the dataset is removed in the data cleaning process. The cleaned
dataset is send to the data visualization process. In data mining process the data’s are analyzed
with different factors like Artist, Title, Album, Length, Year, Genre, Rating, Bitrate, Path and
media of the songs.
38

In the data analysis process, C graph analysis method is used to find the album with
highest preference. Descriptive analysis method is used to find the average life cycle of the hit
songs. To find the most played album in each dataset, pie chart is used and it is found by the
count of each album. The changes of tempo between the starting and ending of the event is
analyzed with the bit rate of the album and it is represented in the line graph. To find the most
played track in each continent, pie chart is used with track list and the correlation between the
music data and its quality is found for each dataset. Finally all the results are compared with each
other to find the answers for the research questions.
Evaluation
RO1 requires to know how the preferences changes over time, this answer is evaluated
and the results denote that Providence US has highest preference. Ratings of the song and Album
names are used to find the song’s preference. The preference is evaluated based on the tracks
played and high rated scores, in all the locations. When compared to other places, Providence
US’s tracks were rated more. i.e., 100 ratings were provided for 30 albums. The other albums
were rated 100 only for less albums.
RO2’s answer is evaluated, where Providence US is determined as the region which
probably begins new trend. RO2’s answer is identified based on RO1, where Providence US is
determined to have highest preference and it is obvious that large crowd will be available in this
region. Thus, new trends are possible where more number of crowd is available.
RO3 is evaluated that in 2015, in Florida Keys US, the Tango Marathon of Friday
afternoon reached average lifecycle of a hit. It was determined based on the means of the ratings,
where descriptive statistics analysis is used to determine the result. The mean value estimated are
as follows- For London’s music rating the mean = 52.68; for Miami US’s music rating the mean
= 67.78; for Providence US’s music rating the mean =82.75; for Brussels’ music rating the mean
= 92.29 and for Florida Keys US’s music rating the mean = 98.18. Therefore, the highest mean
value is for the music rating of Florida Keys US (i.e., mean = 98.18).
RO4 is evaluated, and the China Harbor 2 is the highly played orchestras among all the
continents. This answer was evaluated based on the pie chart, using G-graph statistical method.
The mostly repeated albums are determined for this result.
RO5 is evaluated where the highest dance tempo impact is found for Instrumentales
Involvidables. The following bit rate determination of all the five locations helped to find the
39

answer, where London had highest bit rate for the albums named 1940-1950 and Alex for
Shorey, with 400 bit rate. Miami US had highest bit rate for the album named la Segunda
Tradicion, with 430 bit rate. Providence US had highest bit rate for the albums named Roberto
Maida Canta and Robin's Canaro, with 310 bit rate. Brussels had highest bit rate for the album
named Instrumentales involvidables, with 500 bit rate. Florida Keys US had highest bit rate for
the album named Los Provincianos-1931/1934 Trio Cinaco Ortiz-1931/1955, with 420 bit rate.
Thus, among all Instrumentales Inolvidables has the highest bit rate.
RO6 is evaluated to find the most often or highly played music tracks in all the continents
and the answer is a title track Instrumentales Involvidables in Tierra Querida album by artist
Osvaldo Pugliese. This track has scored 100 ratings, with 500 bitrate. The evaluation is
determined by comparing the ratings of tracks in RO7, where the other continents have the
following tracks with highest ratings.
In London, Ricardo Tanturi con Alberto Castillo’s album 1940-1950 with a title track
Pocas Palabras has scored 100 ratings and 397 bit rate.
In Miami US, Ciriaco Ortiz’s album Los Provincianos-1931/1934 Trio Ciriaco Ortiz-
1931/1955 with a title track Un Placer has scored 100 ratings and 441 bit rate.
In Providence US, Francisco Canaro’s album Robin's Canaro with a title track Le Bien
Pager has scored 100 ratings and 320 bit rate.
In Brussels, Osvaldo Pugliese’s album Tierra Querida with a title track Instrumentales
Inolvidables has scored 100 ratings and 500 bit rate. In Florida Keys US, Ciriaco Ortiz’s album
Los Provincianos-1931/1934 Trio Ciriaco Ortiz-1931/1955 with a title track Un Placer
has scored 100 ratings and 441 bit rate.
RO8 has determined the similarities and differences in the track selection that, are sent
from different DJs of different parts of the world. The similarities include the ratings for all the
tracks i.e., all the tracks have 100 ratings as they are the most popular songs selected from the
selection. The differences include the bit rate is differing for all the selected tracks. The bit rates
are397, 441, 320, 500 and 441respectively for the continents London, Miami US, Providence
US, Brussels and Florida Keys US.
40

1.1 Validity of Results
Validity of the research encompasses the whole data analysis process and establishes whether
the obtained results meets the requirements of the research questions. Validity of the results is
more important for the research. It indicates the importance of our research. This validity can be
applied in both design and method of the research. It controls all possible factors which affects
the research. It is the primary factor for all the researcher. There are two types in validity such as
internal validity and external validity.
Internal Validity
Internal validity is affected by the mistakes within the study itself. It controls some major
variables of the research such as problems or design issues in the research. The internal validity
of the research is affected by the following factors.
 Variables in the dataset.
 Size of the dataset
 Time given for data gathering or research
 Sensitivity of the tasks
 History
 Maturation
External Validity
External validity is used to extent the findings to other contexts or groups. If the external validity
of the research is affected then the findings of this research cannot be applied inn other contexts.
Hence it is important to conduct a research in a highly controlled environment. The factors
which affects the external validity of the research is given below.
 Characteristics of the dataset
 Selection of dataset
 Explicitness of the independent variables
 Effects of analysis or researcher
 Data collection methods
 Effect of Time
To ensure the internal validity of the research, the collected data set is preprocessed in the data
mining process and the issues, data inaccuracy, missing of data are resolved in the data cleaning
process. The size of the dataset, time required for the data collection and sensitivity of the data
41

analysis process is checked and issues in those process are removed to increase the internal
validity of the research. To ensure the external validity of the dataset, simplest data collection
methods are used, the selection of data set, characteristic of the dataset are analyzed and the
explicitness of each variables is checked and the time is used efficiently to increase the external
validity of the research.
The results which are obtained from data analysis process using different types of dataset
are analyzed with different factors such as Ratings of the album, average number of hits, mostly
played album and tracks in each dataset, Bit rate of the album and correlation between the music
data and its quality. To validate the obtained results they are compared with each other to
validate each dataset.
1.2 Generalization of Results
From the overall generalization of result analysis, one playlist is selected. The tracks are
as follows:
1) Pocas Palabras
2) Un Placer
3) Le Bien Pager
4) Instrumentales Inolvidables
The generalization has taken place by the following data:
According to the data set of London it is observed that, Ricardo Tanturi con Alberto
Castillo’s album 1940-1950 with a title track Pocas Palabras has scored 100 ratings and 397 bit
rate. Whereas based on the data set of Miami US, Ciriaco Ortiz’s album Los Provincianos-
1931/1934 Trio Ciriaco Ortiz-1931/1955 with a title track Un Placer has scored 100 ratings and
441 bit rate. Then, the data set of Providence US shows that, Francisco Canaro’s album Robin's
Canaro with a title track Le Bien Pager has scored 100 ratings and 320 bit rate. As per the dataset
of Brussels the results show that, Osvaldo Pugliese’s album Tierra Querida with a title track
Instrumentales Inolvidables has scored 100 ratings and 500 bit rate. On the other hand, from the
data set of Florida Keys US, the results represent that Ciriaco Ortiz’s album Los Provincianos-
1931/1934 Trio Ciriaco Ortiz-1931/1955 with a title track Un Placer has scored 100 ratings and
441 bit rate.
42

2. Evaluation, Reflections and Conclusion
This chapter evaluates the important research aspects by certain reflections that state the
things that works effectively and the things which went wrong. To evaluate this all the phases of
the research are considered.
2.1 Literature Review
According to (Khan and Shah Khan, 2011), the presence of visualization methods can be
quite confusing in understanding the necessary technique to be utilized. But, the authors states
that visual representation is mainly used for effective interpretation of data insight. Visualization
is defined as a mental image/ visual representation of any required instance or object, as a visual
perception. This research papers represents various techniques for visualization, for helping the
other budding researchers to understand about visualization. Further, the authors confirms that
visualization has both advantages and disadvantages, then the problems of visualization are
categorized based on unique perspective, fundamental interactive methods, commonly known
data, information visualization methods, interactivity process, and its scope. The problems
highlighted in this research includes- scalability, usability, Education and Training,
understanding elementary perceptual cognitive tasks, then it also lacks intrinsic quality measures
and other issues.
As per (M. Gustafsson, K. Gustafsson and E. Hammarqvist, 2007), visual data
exploration helps to have instant data exploration and supports to have effective results, when
compared to automatic data mining algorithms. The visual data mining methodology comprises
the data type to be visualized, to have visualization method, and requires both interaction and
distortion. The authors state that visualizing data mining models are trustable and quite helpful
to understand.
2.2 Reflection on Topic and Objectives
The research objectives are met, where the measurable factors for the selection of music
playlist is completed. For visualization, music data is an interesting platform. My personal
opinion on visualization can help to improvise the data for conveying and to present to the
analysts for easing their work. On the other hand, it has the flexibility to visually recognize and
manipulate the data, by addition and removal of new insights on the data. Tango events are very
43

popular and it includes contribution of various DJs from different parts of the world, for
gathering the music track playlist/ data. In contains various events and with different crowd.
Satisfaction of the crowd, by entertaining them is the ultimate goal of this event. If the crowd is
not satisfied then it can decline the popularity of the tango events. Thus, music data visualization
is a challenging task for the researcher. It is difficult to select a specific topic for the research.
This research is a real-time project, which requires challenging research approach, to resolve the
problem. The challenges interests other analysts, researchers and academic supervisors. This
research has originality and this topic is effective in presenting the efficiency of the
methodology, for measuring and implementing the research’s accomplishments. The objectives
of the research were determined clearly, in the beginning stage.
2.3 Reflection on the Selected Methods
The statistical Methodology is the guideline utilized for executing this project. This
method includes repetitions in data analysis, which had great impact on determining the research
objective’s results. This real world problem is analyzed successfully and it is observed that
virtualization method has helped a lot. Visual Analytics has supported in analytical reasoning.
Visual Data Mining comprises of Data Cleaning, Data Transforming and data preprocessing,
which supports in data structuring. Thus, visual analytics has ensured to take right decision with
respect to flexibility. Henceforth, Visual Analytics and Visual Data Mining supports this
research to have knowledge about the acquired data. The accomplished results also proves the
usage of visual data mining that, it has effective data insight.
2.4 Reflection on the Plan
There was difficulty in finding the data, because this is an original research work and
more research is not carried out. It created issues in data analysis and increased the problems for
determining the results for the research objective. The data cleaning was not provided in a
specific format, while collecting data from the DJs around the world. To overcome this problem,
text mining method is utilized and the problem is resolved with effective data separation, with a
particular format for the whole dataset.
44

2.5 Accomplishment
The music industry feels a sigh of relief, as this research presents them a new way to
select the music playlist for any events. This study helps in producing playlist that has
appropriate songs, based on the factors like seasonal trend and location.
The data is collected, and data cleaning is achieved. The gather playlists from the DJs
around the world are amended to have similar format by recovering the missing elements. The
gathered data is visualized, then with Data Visualisation and SPSS software data analysis is
performed. The TF-IDF method is efficiently utilized. All the research answers are answered, by
combining the elements and its frequencies i.e., the question related to preferences change over
time is answered; the region which will probably begin with a new trend is found; the duration
for an average lifecycle of a hit is determined; the places where most often the orchestras are
played are identified; The impact of dance tempo is evaluated; often or highly played music
tracks in all the continents are successfully recognized and compared; the pattern for music track
selection is determined; and the resemblance or differences of track selection, from different DJs
of different parts of the world are investigated. Providence US is determined as the region which
probably begins new trend.
China Harbor 2 is the highly played orchestras among all the continents. It was
determined depending on the pie chart, using G-graph statistical method. The highest dance
tempo impact is found for Instrumentales Involvidables. Using the mean, it was determined that,
in 2015, in Florida Keys US, the Tango Marathon of Friday afternoon reached average lifecycle
of a hit.
The title track Instrumentales Involvidables in Tierra Querida album by artist Osvaldo
Pugliese is determined as the most often or highly played music tracks in all the continents. Due
to its high rating i.e., 100 and high bit rate as 500.
The generalized result has resulted in a playlist with the following tracks:
1) Pocas Palabras
2) Un Placer
3) Le Bien Pager
4) Instrumentales Inolvidables
45

2.6 Conclusion
The current research is completed successfully by collaboration of data, using the SPSS
software. The project accomplished to deliver the required results, where effective visualization
prototype was designed which projects the data in easily understandable picture. The process
workflow includes the functionalities of visual data mining, for enhancing the current process.
The data of playlist, from the DJs of various parts of the world for Tango events are gathered.
The research objects are determined and crosschecked to meet its requirements. Visualization is
completed for music data. The results are represented and discussed. The risks are evaluated. The
data mining process is used, where SPSS software helps in data analysis.
Further, the track playlists from different DJs present in various parts of the world are
collected; data processing using data analysis is performed to find the answers of the research
questions; the research beneficiaries are identified; the limitations are also identified;
visualization solution is designed and implemented; and the results are evaluated and discussed.
The highest dance tempo impact is found for Instrumentales Involvidables. It is observed that
with the help of
mean, in 2015, in Florida Keys US, the Tango Marathon of Friday afternoon has reached average
lifecycle of a hit. New trend is believed to begin in Providence US, due to the track’s popularity
and based on the huge crowd.
China Harbor 2 is the highly played orchestras among all the continents. It was
determined depending on the pie chart, using G-graph statistical method.
It is determined that a title track Instrumentales Involvidables in Tierra Querida album by
artist Osvaldo Pugliese is the most often or highly played music tracks in all the continents.
Because, this track has scored 100 ratings, with 500 bitrate.
A playlist is selected as the best playlist for the tango event, which contains the following
tracks,
1) Pocas Palabras
2) Un Placer
3) Le Bien Pager
4) Instrumentales Inolvidables
46

2.7 Future work
It is observed that the visualization method can do wonders for the music industry. The
current prototype was designed for selection of effective music track playlist for the tango
events, which helps the DJs to keep the crowd engaged in enjoying the whole event, irrespective
of location and seasonal trend. The future work would be to enhance the developed method that
can provide effective data collection with a specific format and even additional features can be
implemented.
References
M. Gustafsson, B., K. Gustafsson, J. and E. Hammarqvist, R. (2007). The Role of Visualization
in Data Mining. University of Linköping.
Blascheck, T., John, M., Kurzhals, K., Koch, S. and Ertl, T. (2016). VA2: A Visual Analytics
Approach for Evaluating Visual Analytics Applications. IEEE Transactions on Visualization
and Computer Graphics, 22(1), pp.61-70.
Health.herts.ac.uk. (2017). QUALITATIVE DATA ANALYSIS. [online] Available at:
http://www.health.herts.ac.uk/immunology/Web%20programme%20-
%20Researchhealthprofessionals/qualitative_data_analysis1.htm [Accessed 16 Dec. 2017].
Information visualization and visual analytics: challenges and opportunities. (2013). SCIENTIA
SINICA Informationis.
Qualitativedataanalysis.net. (2017). Expert Approaches to SPSS Qualitative Data Analysis
Online | Qualitative Data Analysis. [online] Available at:
http://www.qualitativedataanalysis.net/approaches-to-spss-qualitative-data-analysis/ [Accessed
16 Dec. 2017].
Robertson, J. and Kaptein, M. (2016). Modern statistical methods for HCI.
Ruf, R. and Schwab, W. (2016). Visual Analytics. Controlling, 28(8-9), pp.495-501.
47

SearchBusinessAnalytics. (2017). What is visual analytics? - Definition from WhatIs.com.
[online] Available at: http://searchbusinessanalytics.techtarget.com/definition/visual-analytics
[Accessed 16 Dec. 2017].
Sindiy, O., Litomisky, K., Davidoff, S. and Dekens, F. (2013). Introduction to Information
Visualization (InfoVis) Techniques for Model-Based Systems Engineering. Procedia Computer
Science, 16, pp.49-58.
Stolper, C., Perer, A. and Gotz, D. (2014). Progressive Visual Analytics: User-Driven Visual
Exploration of In-Progress Analytics. IEEE Transactions on Visualization and Computer
Graphics, 20(12), pp.1653-1662.
Techopedia.com. (2017). What are Visual Analytics? - Definition from Techopedia. [online]
Available at: https://www.techopedia.com/definition/30371/visual-analytics [Accessed 16 Dec.
2017].
Vismaster.eu. (2017). Chapter 2 – Visual Analytics - VisMaster. [online] Available at:
http://www.vismaster.eu/book/chapter-2-visual-analytics/ [Accessed 16 Dec. 2017].
ZenTut. (2017). Data Mining Techniques. [online] Available at: http://www.zentut.com/data-
mining/data-mining-techniques/ [Accessed 16 Dec. 2017].
Keim, D. (2002). Information visualization and visual data mining. IEEE Transactions on
Visualization and Computer Graphics, 8(1), pp.1-8.
Yu, C., Yurovsky, D. and Xu, T. (2011). Visual Data Mining: An Exploratory Approach to
Analyzing Temporal Patterns of Eye Movements. Infancy, 17(1), pp.33-60.
Khan, M. and Shah Khan, S. (2011). Data and Information Visualization Methods,
and Interactive Mechanisms: A Survey. International Journal of Computer
Applications (0975 – 8887), 34(1).
48