Techniques of Predictive Analysis

This literature review explores the applications of predictive analysis in different industries, such as medicine and business. It discusses the use of data mining techniques and the importance of accurate predictions for strategic decision making.

Techniques of predictive analysis
Techniques of Predictive Analysis
Student Name
Institution
City, Country
Email
Abstract-The world of business today is
filled with magnitudes of data, the
quantity of data being collected by
various industries is continuously
flourishing as digital technology continue
to penetrate the industries. The increased
exploitation of a number of novel
inventions and the impact of social media
has availed a large data set termed as big
data. This data has the capacity to
generate very effective information if
appropriate models are applied to analyse
it. Organization that were previously in
the verge of collapse can instantly rise to
success curtesy of the big data technology,
for this reason several industries are
increasingly paying attention to the
various predictive models that can be
applied to assist forecast future states of
nature. The predictive analysis composed
of a number of statistical and analytical
techniques that can be applied to develop
models for predicting future possibilities.
For this reason, predictive analysis has
proved to be a vital area in cases where
substantial quantity of sensitive data
needs to be analysed. With the aid of the
data mining techniques, future
probabilities and measures can be
forecasted and the predictions used to
improve strategic decision making. This
literature review identifies clear ideas of
the applications of the data mining
techniques and the sue of the predictive
analytics on different areas such as
medical field and business in particular.
I. Literature review
A. Introduction
Predictive analytics is the science
and art of developing predictive models
which can afterwards be applied to predict
selected outcomes in the future with a higher
probability of occurrence than would be
under a normal gauze. Predictive analysis is
often applied as an umbrella term that also
entails other forms of advanced analytics
such as the descriptive analytics that focuses
on giving an insight of past occurrences. The
concept of predictive analytics applies large
and varying techniques to assist firms
foresee the future. As the idea of big data
continues to take over business strategic
decision making, the application of
predictive analysis is continuously becoming
popular in firms [1]. Technologies like
machine learning, txt analysis as well as
neural networking are some of the
applications of predictive analysis in
business.
The application of big data has gained
popularity in most of the organisations
operating globally, through the use of big
data firms are relying on the available
information to evaluate expected consumer
reactions, gauge potential products demand
as well as analysis of the financial impact of
management decisions. The predictive
analysis techniques apply technology to

Techniques of predictive analysis
forecast the future occurrence and enable the
managers to act accordingly. The ability to
foresee the future is a very significance
concept in a business. Through accurate
predictions firms are able to take steps that
will improve their future competitive
advantage. Global business is increasing
becoming competitive, the techniques of
predictive analysis are thus a significance
avenue for meandering the competitive
business world. The ability to predict the
future do eventually define the firms that
comes out on top in the long run.
B. Summary of articles
The concept of predictive analytics has
been in the past applied in the area of data
mining to forecast future events especially in
the field of medicine, education as well as
crime detection. The health sector does
contain bulky information that if used
effectively can be significant in making
crucial decisions. The research conducted by
Babu and Sastry [1] was focused on the
enterprise resource planning (ERP)
predictive features. The study objective was
on how the current and past data can be
analysed and applied to identify probable
risks that organisations might be facing.
Analytical decision tools are utilised by the
system to make decisions as a way of
improving service delivery. A study by
Bellaachia and Guven [2] did suggest that
data mining can be efficiently applied to
forecast lastingness of breast cancer. The
authors did examine three methods applied
in data mining (neural networks, Bayes and
Naïve) to show that data mining is the apt
technology that can identify patterns in a
data set derived from the health sector. Even
though some diseases are more difficult to
predict due to their complexity the
application of predictive together with
advanced skill in the medical field can assist
identify some trends that may assist lead to
breakthrough. To study heart diseases by
applying the classification algorithms, data
mining algorithms like naïve Bayes can be
useful in forecasting heart attacks. From the
author’s analysis the use of this tools can
give up to a 99% accurate prediction [3].
The data mining techniques did prove that
diagnoses can be predicted when the right
techniques of predictive analysis are applied
in the right manner.
A survey conducted regarding
predictive analysis by using big data did
show that large volumes of medical data can
only be processed accurately by applying
very powerful analytical tools. Techniques
under data mining can also be applied in the
analysis of trend in diseases. Ramaraj and
Thanamani [4] proposed that heart diseases
can be significant identified by the use of
predictive analytics. The objective of the
researchers was to design a predictive
method that can be applied in the detection
of heart diseases. From their conclusion, the
use of CN2 rule is best fitted for doing
classifications compared to any other
technique.
A study by Nasridinov et al [5] did
analyse several data mining with generated
test data meant to evaluate the most efficient
technique to predict patterns in crimes. The
researches did focus mostly on the extensive
performance analysis of several data mining
algorithms. The study was based on the
assumption that wearable sensors are
attached to the clothes of the users of the
identified techniques. The sensor would
capture the user’s inner temperature and
heartbeat and afterwards send the
information to the servers to perform
emotion miming. Danger was indicated wen

Techniques of predictive analysis
the user did develop faster heartbeats and
high inner temperature. Once the danger was
detected the researcher would apply a test
data generation method that precisely design
prediction algorithms. This technique can be
applied by law enforcement bodies and
emergency service teams to predict trends
and abstract useful information that can
assist curtail several danger situations
promptly.
The use of case-based machine was
efficiently applied by Chandra et al [6] to
develop a more accurate algorithm for
forecasting heart diseases. The method was
useful when applied to binary dataset. The
use of non-binary data presents a number of
challenges as it initially requires innovative
methods to calculate support. The presence
of a chance to remove a candidate item from
the non-binary dataset due to pruning and its
application to a higher level raises the
frequency of pruning occurring. The final
occlusion by the author was a prototype that
was meant to develop request item sets
needed for developed set of non-binary data.
A study conducted by Kone and Karwan
[7] which was designed to predict the
expense of delivering liquefied gas to new
clients by using multifactor regression
model did indicate that evaluating all the
observations at once did lead to a poor
prediction of the possible outcomes. For this
reason, before undertaking the regression
analysis there was need to utilise a novel
supervised leaning technique to group
outcomes with similarities in certain
perceptions. Hyperboles are applied to
devote customer classes and afterwards a
linear regression is developed taking
account of the classes. Increasing the
culmination of the data classification and
regression did indicate the accuracy of the
predictions made. The application of
regression for predictive analytics was also
emphasized by Bhat et al when they
presented a processing phase that was
attached to the imputation of missing values
for both numerical and categorical data.
Classification and regression trees together
with generic algorithms were suggested
forefooting the missing values and self-
organisation of the feature maps so as to
enable impute of categorical values applied
in the research.
Jakrarin and Piromsopa [9] in their study
aimed at efficient analysis of a magnitude of
data and reduction of time complexity,
applied the k-means clustering which made
use of the Apache Hadoop technology. They
were also interested in identifying whether
the detection and accuracy rates are
influenced by the number of fields that are
present in a log file. Given that accuracy is
reduced when the number of entries goes up,
the understanding from the research did
indicate that there is need to improve the
accuracy. Another technique was proposed
by Jun et al [10] termed as divided
regression analysis. This system was
suggested to be effective in big data analysis
to assist minimize the computing burden that
is experienced in regression problems. The
divided regression technique is a statistical
method that is focused majorly on small data
samples hence reduce the burden of
computations that is experienced in the big
data analytics. The method is recommended
to be applied in the analysis of the entire
data set in big data analysis which is referred
to as the population set in statistics.
C. Conclusion
Predictive analysis is a vital branch of
data engineering that is usually focused on

Techniques of predictive analysis
the prediction of existence or a probability
of data. Predictive analytics applies data
mining techniques to come up with
predictions about events that may occur in
future. The forecasts can thereafter be used
to develop recommendations that can be
applied to the decision-making process. The
procedures under predictive analysis is
composed of analysis of past data which is
thereafter applied in foresting the future
occurrence. Two major objectives of
predictive analytics are classification and
regression. The analytics is composed of
different analytical and statistical techniques
that are applied for evolving the models
meant to und3rtake future prediction of
events. The predictive analysis has the
potentiality to deal with both continuous and
discontinuous changes. Classification
prediction do comprise of analytical
techniques that are applied in the predictive
analytics. The roles of the predictive models
do vary depending on the data that is used
together with them. In this literature review,
we have identified past research work and
explained how the researchers applied
predictive analysis to solve problems in the
medical and the business world. In addition,
the review has identified some of the
techniques and algorithms that were
implemented while applying the big data
technology. This information does indicate
that it is possible to develop a model that
could assist apply predictive analytics
modelling techniques that is significance for
synthesis of big data.
II. References
[1] P. Selvaraj and P. Marudappa, "A survey of predictive analytics using big data with
data mining," International Journal of Bioinformatics Research and Applications, vol.

Techniques of predictive analysis
14, no. 3, p. 269, 2018.
[2] P. Babu and S. Sastry, "Big data and predictive analytics in ERP systems for
automating decision making process," 5th IEEE International Conference on
Software Engineering and Service Science (ICSESS), p. 259–262, 2014.
[3] A. Bellaachia and E. Guven, "Predicting breast cancer survivability using data mining
techniques," Department of Computer Science, the George Washington University,
Washington , 2005.
[4] H. Masethe and M. M.A, "Prediction of heart disease using classification algorithms,"
Proceedings of the World Congress on Engineering and Computer Science, San
Francisco, USA, 2014.
[5] M. Ramaraj and A. Thanamani, "Comparative study of CN2 rule and SVM Algorithm
and prediction of heart disease datasets using clustering algorithms," Department of
Computer Science, NGM College, Pollachi, India, 2013.
[6] A. Nasridinov, J. Byun, N. Um and H. Shin, "A study on danger pattern prediction
using data mining techniques," School of Computer Engineering, Dongguk
University, Gyeongju, South Korea, 2014.
[7] K. Chandra Shekar, K. Ravi Kanth and S. K., "Improved algorithm for prediction of
heart disease using case based reasoning technique on non-binary datasets,"
International Journal of Research in Computer and Communication Technology, vol.
1, no. 7, p. 420–424., 2012.
[8] E. Kone and M. Karwan, "Combining a new data classification technique and
regression analysis to predict the cost-to-serve new customers," Computers &
Industrial Engineering, vol. 61, no. 1, p. 184–197, 2011.
[9] T. Jakrarin and K. Piromsopa, "An analysis of suitable parameters for efficiently
applying K-means clustering to large TCP dump data set using Hadoop framework,"
Electrical Engineering/Electronics, Computer, Telecommunications and Information
Technology (ECTICON), 10th International Conference, Krabi, Thailand, 2013.
[10] S. Jun, S. Lee and J. Ryu, "A divided regression analysis for big data," International
Journal of Software Engineering and its Applications, vol. 9, no. 5, p. 21–32. , 2015.