A Technical Seminar Report on

TITLE OF THE TOPIC

Submitted to
The Department of Computer Science and Engineering
Bachelor of Technology
In

Computer Science and Engineering

(2018 – 2022)

K.SRIYA
19R11A0519

Department of Computer Science and Engineering

Geethanjali College of Engineering and Technology (UGC Autonomous)
Accredited by NBA, (Affiliated to J.N.T.U.H, Approved by AICTE, New Delhi)
Cheeryal (V), Keesara (M), Medchal.Dist.-501 301.
May-2022
 Department of Computer Science and Engineering
Geethanjali College of Engineering and Technology

CERTIFICATE

This is to certify that this technical report on “PREDICTIVE ANALYT AND

ITS IMPLEMENTATION IN CYBER SECURITY”, submitted by
MATTAPALLY SAHITHI(16R11A05U3), in the year 2022 in partial
fulfillment of the academic requirements of Geethanjali college of Engineering
and Technology for the award of the degree of Bachelor of Technology in
Computer Science Engineering, is a bonafide work that has been carried out
by them as part of their Technical presentation during fourth Year Second
Semester. This report has not been submitted to any other institute or university
for the award of any degree.

Date of seminar: 02/03/2020

Coordinator Head of the Department

Faculty Name Dr. A.SreeLakshmi
Designation, CSE HOD,CSE
 ABSTRACT
Cisco packet tracer enables developers to view the flow of data packets and carry out analysis
on the data packets transmitted in the IoT network. All the IoT devices on Cisco Packet
Tracer can be run on standard programs or can be customized by programming them with
Java, Python or Blockly. This work acts as a emergency help which on detecting smoke
above a particular threshold , blows off a siren and turns on the fire sprinkler if the smoke
still persists. It also controls the appliances based on a temperature sensor which contributes
towards building a smart home.
 INDEX

S.NO LIST OF CONTENTS PAGE NO

1 4-6
TECHNICAL REPORT

2 PAPER PRESENTATION SLIDES 7-15

3 BASE PAPER 28
 TECHNICAL REPORT
PREDICTIVE ANALYTICS AS A SECURITY MANAGEMENT TOOL
AND ITS APPLICATION IN CYBER SECURITY
1. INTRODUCTION TO PREDICTIVE ANALYTICS:
The term Predictive analytics is defined as the art and discipline of utilizing data which helps
in discovering relationships and hidden patterns within the data.It doesn't tell what exactly
happens in the future, but forecasts what will happen and provides valued actionable visions
for assured decision-making[3].This technology has grown a lot in these days from
developing a predictive model that has the capability to impact a decision which plays a
major role in cooperating these models into daily operations.[5] The software applications of
predictive analytics use variables that can be measured and analyzed. These variables are
used to forecast the expected performance of machinery, persons or other entities.

Big data, Predictive analytics has also gained a lot of importance and prominence. The usage
of wide number data application tools like Hadoop and other big data platforms have created
a large number of data mining opportunities to increase predictive insights. Amplified
development and profit orientation of machine learning tools by IT merchants has also aided
enlarged predictive analytics proficiencies [9].

2. FEATURESOF PREDICTIVE ANALYTICS:

The following are some of the main features of predictive analytics.

2. 1. New theory development:
This is a significant feature, as the rapid varying situations bring out the huge number of data
sets, which contains numerous types like a digital type of data, text, and the outrageous speed
at which the data is received[2].All these conditions bring about the necessity for fruitful new
theory development. As the definition of Predictive analytics suggests that It helps in solving
many hypothetical situations and predicts the feature. One of the toughest tasks is identifying
relationship sand patterns within the data.for example, consider the data that consists of
relationships and patterns with theories which don’t contain any newly measurable concepts.
Predictive analytics notices these new patterns and behaviors that help in discover new casual
mechanisms and lead to new theory development.[6]
2. 2. Construct operationalization:
This is the key feature of predictive analytics which deals with growth of new theory that
operates concurrently with the idea which had been developed through new theory
development. It develops a new model by comparing diverse operational limitations of
numerous models using assessment methods. It provides construction authentication of the
compared models.[3]

2. 3. Comparing existing theories:

Comparing models using normal methods for calculating predictive accuracy is difficult as
the result may not be to up to the mark[9].Predictive analytics makes usage of explanatory
models for making comparisons among many foremost theories, which are generally difficult
to compare via statistically for calculating predictive accuracy.

2.4. Improving existing explanatory models:

After comparing the models with different explanatory models, we get some verdict
regarding the result obtained via comparison [3]. If there is any situation to improve the
existing model, Improvement can be done by capturing Intricate relationships and patterns of
the models then spinning them into predictive values.

2.5. Assessing the predictability of existing models:

Predictive analytics can be utilized to calculate the foreseeable level of calculable
circumstance by generating a standard point for foretelling correctness, which leads to
enlargement of data collection, new empirical, new methods, approaches and appraisal of
existing models for predictive control [4].

3. PROPOSEDFRAMEWORK PROCESS:
The below fig explains how the proposed framework works and the important stages in it.

3.1. Proposed framework process

3.2. Data Collection:
The initial stage of this process is the collection of data, which is done by using the
application programming Interface (API) and the applied monitoring tools.[2][7] Initially,
predictive goal is established as a standard point. The data which is collected will be
examined and calculated using data-driven algorithms and supervised learning to check
whether it equals with the established predictive goals or not. The data is measured in such a
way that the patterns and relationships of the data which occur in both valid output and input
data produce a predictive value which will be utilized in calculation during model
development or design.[11]

3. 3.Data Preparation:
The collected data from the data Collection stage is prepared and transformed into the related
format and this collected data is used for analysis. This data may contain missing values and
these missing values are identified by using Algorithms of classification and regression. At
this phase, variables and observations will either be removed or added to the data set. The
variables and observations help in determining whether the missing values are needed to be
add to predict the observations or for model training.[1][10]

3.4. Model Design and Development

After proper preparation of data using different algorithms and nullifying the missing values
of the data set, we can design a proper model. After designing the model Unknown or least
expressed relationships of the data sets will be captured using Explanatory data analysis. This
utilizes exploratory techniques to define the capacity of the predictive associations and values
that are at center of predictive modeling.

3.5. Model Evaluation and Testing:

The data which is acquired in the data preparation stage will be measured and predictive
accuracy will be calculated for proof to run predictions[9]. The training data will be
compared with the output data, which helps in evaluating the model for the data obtained in
the initial stages of the process.

3.6. Model Deployment:

The model which had been developed using this process will be deployed and that model will
be shaped to examine the theory. This includes observing and examining the model for
calculation of predictive accuracy.
4. COMMON TYPES OF SECURITY THREATS:
4. 1. Phishing attacks:
Basically, Phishing email attacks are the type of attacks that are untargeted. In this scenario,
the large number of emails will be sent to several receivers which are meant to obtain delicate
information (e.g. username, password, or bank details)[4][3]. The obtained information helps
a hacker to accomplish a data rupture. It’s tough to identify a phishing email because it is
hidden as a reliable entity that stimulates a recipient to visit websites, whose intention is to do
harm.

4.2. Insider security threats:

The evolving methods which are common in many internal security threats and data breaches
are composed by your specific employees. Inside extortions can be malicious or
unintentional. Irrespective of a data rupture from the inside is mainly tough to sense.

4. 3. Denial of service (DoS):

When a hacker overflows a website with extra movements than it can usually maintain then
these type of cyber attacks comes into the picture.[5] As a result, genuine users will not be
able to ingress to services which result in expensive interruption of users or employees
Distributed denial of service (DDoS) attacks uses a botnet – a collection of computers or IoT
devices. Botnets appears like genuine traffic which makes it tough to identify the difference
between malicious and normal traffic.

4. 4. Malware:
Malware is one among the well-known security extortions, It refers to several types of
innocuous software executed when a user erroneously downloads it[6]. Malware is involved
in many recent cyber security breaches which also, includes the ransomware attacks
Petya/Not Petya and Winery. Old-fashioned malware such as back doors and Trojans viruses
also remain determined.

5. PREDICTIVE ANALYTICS IN CYBER SECURITY:

5.1. Forewarned is Forearmed:
Cyber security needs a better approach to deal with threats. You need to crunch your data,
identify patterns among the data sets and detect anomalies by which you can close the locks
before your data is filched[9][7].Predictive analytics will give info only regarding where the
cybercriminals are likely to attack next, but not about the place where they have tried to
attack in the past. It intimates about the region where your weak and also foretells how
thrivingly you areready to counter an attack before it is late.

5.2. It Can Handle Huge Volumes of Data:

One of the major confronts for a tough cyber security solution is that you're dealing with
huge pools of data, which can be very rigid to walkthrough, analyze and process for
beneficial visions. These data streams come from a wide range of programs, databases, and
devices, which tells you that you need a prevailing BI tool to keep them all in sync.
Predictive analytics helps your system in thriving on huge pools of data. In fact, If you work
with a more number of inputs, you get the complete picture– and the more accurate result you
can make with your predictions.[9]

5.3. Automates Much of the Workload:

Basic usage of machine learning and Predictive analytics is to handle your data and tease out
critical vision stremendously quickly which takes much of the pressure off your IT
department. You need not to appoint a various number of employees on a single pattern.
Single expertise is good ample to precisely interpret findings and patterns, the work of
collecting and compiling reports is very tough and is done automatically[8]. Through this, the
team can pay less concentration towards getting stucked down in daily queries and
broadcasting and more concentration on their brainpower on dealing with potential threats
and working hard to shield the system.

6. CONCLUSION:
The present evolving technologies with new characteristics had also carried about fresh
security extortions and dangers. It is essential for security specialists to continue in
discovering solutions to this major challenge i.e, never-ending security extortions and
hazards[10][11].Predictive analytics is the technique used by many of the researchers, but
many may not work with this till the end, few of them end up by providing explanatory
solutions which use data mining techniques to investigate security extortions and hazards.The
quickness of identifying the potential threats and finding the solutions to the threats before
they occur is very important[10].
REFERENCES

[1] Y. Zhang, A. Juels, A. Oprea, and M. K. Reiter, "HomeAlone: CoresidencyDetection in the Cloud via Side-
Channel Analysis," pp. 313-328, 2011.
[2] C.-J. Chung, P. Khatkar, T. Xing, J. Lee, and D. Huang, "NICE: Networkintrusion detection and
countermeasure selection in virtual networksystems," Dependable and Secure Computing, IEEE Transactions
on, vol.10, pp. 198-211, 2013.
[3] Y. Low, D. Bickson, J. Gonzalez, C. Guestrin, A. Kyrola, and J. M.Hellerstein, "Distributed GraphLab: a
framework for machine learningand data mining in the cloud," Proceedings of the VLDB Endowment, vol.5, pp.
716-727, 2012.
[4] G. Shmueli and O. Koppius, "Predictive analytics in information systemsresearch," Robert H. Smith School
Research Paper No. RHS, pp. 06-138,2010.
[5] G. Shmueli, "To explain or to predict?," Statistical science, pp. 289-310,2010.
[6] Y. Tan, H. Nguyen, Z. Shen, X. Gu, C. Venkatramani, and D. Rajan,"Prepare: Predictive performance
anomaly prevention for virtualizedcloud systems," in Distributed Computing Systems (ICDCS), 2012 IEEE
32nd International Conference on, 2012, pp. 285-294.
[7] A. Sanfilippo, "Fostering the Creation of Decision Advantage throughModel Integration and Evaluation,"
presented at the Workshop on CurrentIssues in Predictive Approaches to Intelligence and Security
Analytics,Vancouver, BC, Canada, 2010.
[8] C. Elkan, "Predictive analytics and data mining," Retrieved from cseweb.ucsd. edu, 2013.
[9] G. Shmueli, "To Explain or to Predict?," vol. Vol. 25, ed, 2010.
[10] Y. N. Tan, Hiep Shen, Zhiming Gu, Xiaohui Venkatramani, Chitra Rajan,Deepak, "PREPARE: Predictive
Performance Anomaly Prevention forVirtualized Cloud Systems," p. 10, 2012.
[11] P. S. Rahul Singh, Maitreya Natu, Vaishali Sadaphal, Harrick Vin,"Analytical Modeling for What-if
Analysis in Complex Cloud ComputingApplications," Performance Evaluation Review, vol. 40, p. 10, 2013.106
SLIDES
[BASEPAPER]