ABC Multimedia Video Data Analysis: ITECH1103 Group Assignment

Desklib provides past papers and solved assignments. Access this student-submitted ITECH1103 big data analytics project.

ITECH1103- Big Data and Analytics
Group Assignment
ANALYTIC REPORT & PRESENTATION
1

Contents
Background information...................................................................................................3
Reporting Dashboards..................................................................................................... 4
Advance insights of dataset.............................................................................................9
Research........................................................................................................................11
Recommendations......................................................................................................... 13
Cover letter.....................................................................................................................14
Team reflection.............................................................................................................. 15
References.....................................................................................................................16
List of figures
Figure 1: Answer Q1-Q5..................................................................................................4
Figure 2: Answer Q6-10...................................................................................................5
Figure 3: Top 10 most viewed videos country wise..........................................................6
Figure 4: Top 10 least viewed videos country wise..........................................................6
Figure 5: Answer Q11-15.................................................................................................7
Figure 6: Answer Q16-20.................................................................................................7
Figure 7: Advanced insight dashboards.........................................................................11
2

Background information
The company ABC online multimedia company deals in the online multimedia services.
As a Content of the AMC online multimedia company, the task is to explore the selected
dataset in order to better understand, analyze and visualize the data set. For the
analysis and visualization of the data and understanding the content for the company,
the data set that was selected here is of the videos that were uploaded from the year
2006 to 2018. The dataset used here is originally from the Kaggle.com and then it was
modified and used in order to get a better insight into the data and visualize the data for
ABC online multimedia company. The data set that was used here was analyzed and
visualized using the analytics tool of the IBM Watson. IBM Watson is used to processing
and visualizes the dataset using the simple approach. The features of the IBM Watson
allow the user to simply upload the dataset on the system and analyze the data using
the simple set of question (Bowie, 2012).
The dataset selected here had information related to the videos that were uploaded
from the year 2006 to 2016. This data was of four different countries and in various
categories of video. These categories include categories like music, documentary,
movies, education, news & politics, pets & animals, comedy and other (Eysenbach,
et.al., 2016). The data set could be understood using the following data dictionary:
3

Using the data set, the data was analyzed using some 20 questions and some extra
insight versions. After the analyzation of the data, the dashboards and the other graphs
will be constructed to produce the best data analytics report. The insights gained from
the analysis of data using the IBM Watson will be used to suggest a recommendation
for the better improvement and management of the video data.
Reporting Dashboards
The selected dataset was analyzed and visualized using the set of 20 questions.
Figure 1: Answer Q1-Q5
As shown, in the above picture, the answers of the first five questions were obtained
and then integrated together to form a dashboard for the Content Analytics of ABC
online multimedia company.
A1. The total number of videos in the dataset were obtained using the summary of the
video_id option, and the total number of uploaded videos are 55885. These videos were
counted individually.
4

A2. The total number of categories in the dataset were obtained using the summary of
the category_id option, and the total number of upload categories is 18. These
categories were counted individually.
A3. The total number of countries in the data set is equal to 4 and this was obtained
from the summation of the publish_country. The countries are France, Canada, the US,
and GB.
A4. The total number of channels in the dataset is equal to 12360 and this was obtained
from the summation of channel_title, counted individually.
A5. The graph of the channel_title VS publish_country was created and it was found
that the top three countries according to the number of channels are: France, Canada,
and the US (decreasing order).
Figure 2: Answer Q6-10
A6. The channel_title and publish_country were plotted against each other & the lowest
number of channels were in GB as1624 individual channels.
A7. The total no of channels in the US was found to be 2207.
A8. The channels that were viewed the most according to each country are:
5

Figure 3: Top 10 most viewed videos country wise
A9. The channels that were viewed the least according to each country are:
Figure 4: Top 10 least viewed videos country wise
A10. The Year (publish_date) was counted individually and it was found that there are
13 different years in which videos were uploaded, starting from the year 2006 to 2018.
6

Figure 5: Answer Q11-15
A11. In the last month i.e., the last month of the year December there were 8397 videos
that were uploaded. This was counted using the Month (publish_date) and Year
(publish_date).
A12. In the year 2018, the maximum number of videos were uploaded. This was
calculated using the publish_country: GB, and Year (publish_date).
A13. The maximum no of videos was uploaded in the time 04:00-04:59 pm in the
countries France, Canada and US. The upload time for GB varied by an hour, from
05:00-05:59 pm. This was calculated using the time_frame.
A14. The maximum videos were viewed in music, film & animation and spots
categories.
A15. The minimum videos that were viewed belong from peoples &blogs, news &
politics and trailers.
Figure 6: Answer Q16-20
A16. The most liked video in the dataset is 21 Savage, Offset, Metro Boomin - Ric Flair
Drip.
7

A17. The most disliked video in the dataset is #ProudToCreate: Pride 2018.
A18. The most videos were uploaded on the 6th (Friday) and 7th (Saturday) day of the
week.
A19. The least number of videos were uploaded on the 1st day (Monday) of the week.
A20. From the monthly breakdown, it was understood that the maximum videos were
uploaded in the month December & November and the minimum was uploaded in
August and other months.
8

Advance insights of a dataset
For the advanced insight in the dataset the following questions were asked:
AQ1. What was the variation in view to the uploaded videos, from the year 2006-2018?
the image above displays the change in the view received for the videos uploaded.
There was on constant increase or decrease in views. The most views were gathered in
the year 2018.
AQ2. On which day of the week, maximum views were recorded?
From the image above, it is clear that the maximum views were gained on Friday and
minimum on Saturday.
AQ3. In which time frame, maximum and minimum views were gained/
9

From the graph, it is clear that the maximum views were gained in 4:00 to 04:59 and
minimum in 19:00-19:59. This displays the active time of the viewers.
AQ4. Display the trend of likes over publish day?
From the graphs, it is clear that on Friday most likes were gained and on Saturday
minimum likes were gained.
AQ5. Show the comment and likes trend in different categories.
From the image above the maximum likes and comments were observed on the music
category videos and the minimum likes and comments were observed on classics, auto
&vehicles, and trailers. This shows the category preference of the viewer.
10

Research
Figure 7: Advanced insight dashboards
The advanced research in the data was done using the advanced insight questions.
These questions were used in order to analyze the data more deeply and constructively.
This was done to make the research more valid and to bring out more information from
the data to give the company a perfect visualization which can help them in trading their
work of online multimedia handling (Zoltan, 2018).
The variation in the views according to the uploaded videos was analyzed in order to
know how the viewers of the videos changed over time. It was found that there is no
fixed pattern of the viewers. The viewers decreased and increased over the years
(Aristides, 2013).
The variation because in the views received on the videos were analyzed to understand
on which day the most viewers are active. This helps the uploading video
party/channel/group to know that which day will bring them more active viewers, i.e.,
more likes, views and comments (Data analysis and visualization, 1995).
The analysis of the time frame of the view of videos helped the company to understand
that there are fixed times on which the maximum numbers are active and are available
to watch. This will in turn help in understanding the time and day of the week on which
11

the viewers will be actively using the channel and this will help them in gaining profit
(Guidi, Miniati, Mazzola, & Iadanza, 2016).
On Friday, the maximum number of the likes on the videos were gained. This was the
result of the 4th query of the advanced insight and this help in understanding that the
videos uploaded on Friday were liked the most. This is not because of the good content
of the video but the energy of the viewers because of the coming weekend which
automatically puts them in good mood (Pugh, Wickens, Herdener, Clegg, & Smith,
2017).
For understanding the viewer, it is very important for the company to know what kind of
content is most liked and preferred by the viewers. In order to understand that, there is
a need to understand what category video has been more active and what is the
preferences f the viewers. This will help the company to understand that the viewers like
the particular category and dislike the particular set of categories. This allows the
company to manage the content of the video accordingly (Thorstad, & Wolff, 2018).
This overall analysis helped in understanding, the views and the relative factors that
affect the number of views and likes.
12