Volume 7, Issue 5, May – 2022 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165

Personality Classification with Data Mining

Gaurav R. Savant (M.E. P.T. C.S.E.), Dr. G. R. Bamnote (Professor)
Department of Computer science and engineering PRMITR, Badnera

Abstract:- Personality of a person decide whether he can based on the answers given by user that are relating to five
play the role of leader, influence people around, master personality traits. System will compare data stored in the
in communication skills, do collaborative work, able to databasewith the results and automatically predict the user’s
do negotiateiation in business and handle stress. personality. Based on the personality traits of the user,
Personality describes human features that decides how system will provide classes as modest, semi modest,
people deal with world around. personal behaviour confident, somewhat overconfident, overconfident that
features can be used for deciding a person’s personality describes the user’s personality. Personality decides how a
based on his/her personality traits. Narcissist person is person interact with the outer world. To reveal personality of
having features like seeking attention, over self- a person by analysing the behaviour of him is a conventional
importance, lack of empathy for other people around. technique. As this is manual method of personality
Using big five personality traits, whether a person is prediction takes lot of time and resources. Analysing
narcissist can be determine. By knowing the level of personality based on one’s nature was a difficult task and
narcissism with the combination of five personality traits much human effort will be required to do manual analysis.
personality types like modest, semi modest, confident, Also, this did not provide correct results while analysing the
somewhat overconfident, overconfident can be predicted personality of a user from their nature and behaviour. Since
with APC. Data consists of answers to 40 different analysis was done manually, as humans prone to be
questions, along with their scores, which are used to prejudice and generally see the things accordingly affects
evaluate Narcissistic Personalities. Later when user the correctness and thus decreases the accuracy.
answers a questionnaire related with big five personality
traits, ensemble learner does the prediction and system II. LITERATURE SURVEY
will show the personality of user. This learning can
classify/predict user personality based on past Aleksandar Kartelj etc. al. [1] said that reliable
classifications.This system can be use in organizations, approaches can be used to classify the personality in various
agencies where they recruitcandidates depending upon new researches by applying the concept of Automated
their personality features along with technical Personality Classification. Firstly, examined all the possible
knowledge. Thispaperproposes a system which brings solutions and what all improvements can be made to the
out the personality of an candidate. Personality existing problems of Automated Personality Classification.
classification refers to the psychological classification of Then considered the extension of the Automated Personality
various styles of individual. This project deals with the Classification [APC] problem such as the Dynamic APC and
areas wherever it determines the characteristics of how to remove inconsistency in textual data. This entire
someone. It is often useful to classify person research was carried out in the context of social networks
mistreatment temperament classification mistreatment and related data mining mechanisms. Fazel Keshtkar et. al.
information mining approach. During this paper, aim is [2] said that aims of developing methods for modelling
to alter the personality prediction of the users by taking student behavior based on data such as online conversations,
a mental test using 40 questions related to five discussions in class, etc. However, methods like Intelligent
personality traits. The system uses ensemble Tutoring System (ITS) and Educational Data Mining (EDM)
classification algorithmic rule. The analysis is finished used an individual's behavior and personality for analysis
vast knowledge of information in data set and is been purpose. Thus, a system is developed which can be adjusted
compared with the user input. This paper in the main by the user and analyze student’s behavior during their
focuses on multi classification. interaction as well. Nurbiha A Shukora etc. al. [7] have
given the concept of Online learning which became highly
Keywords:- Personality classification, Narcissism, APC. popular because of technological advancement that made it
possible to have discussions even from a distance. Most
I. INTRODUCTION studies that have been conducted report how effective online
learning has helped students to improve their learning power
To identify the personality of a person by observing while assessing the learning process simultaneously. This
their nature is an old technique. This was a manual process kind of discussion can be possible only by applying data
to predict the nature of the person. Data mining is mining technique wherein one can assess the different
prominently used today by companies with a strong experiences of students which they filled online on the basis
consumer focus- retail, financial, communication, and of their log files. However, it is suggested by the results that
marketing organizations. Various ways used to analyse the students should put more hard work to became an excellent
data are surveys, interviews, questionnaires, classroom online learner. [17]A comprehensive investigation of a
activities, shopping website data, social network data about company's ideal clients is known as a customer personality
the user experiences and problems they face. These analysis. Customer personality analysis enables a company
conventional approach requires more time.This system to adapt its product depending on the preferences of its
reveals information portraying the personality of the user, target customers from various customer categories.the main

IJISRT22MAY1968 www.ijisrt.com 1600

Volume 7, Issue 5, May – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
motive for the paper is to find the accuracy of the prediction IV. SYSTEM ARCHITECTURE
of the personality of the customer who is shopping and
improve the research out there using the ensemble
technique.

III. SYSTEM DESIGN

To solve the issues of the existing system a personality

classification system is designed in which some data mining
techniques ae used and machine learning algorithms are
implemented to classify the personalities of user. It is
achieved by using ensemble classifier of different
algorithms like KNN, Logistic regression, Deep forest, MLP
and Support Vector Machine. By using the previousdata,
new techniques can be applied to identify the personality, so
that it solves the issues with the existing system. In this
proposed system, the user has to answer 40 questions which
are relating five personality traits. Here it converts textual
answers to features,1 for ‘yes’ and 2 for ‘no’. Based on the
obtained results system predict the personality labels
asmodest, semi modest, confident, somewhat overconfident,
over confident.This system can be usedin many sectors like
interview, recruitment process, government sectors,
psychometric tests. Once a user personality is revealed then
he/she can be hired in any organizationwhere they are
allocated with their personality type jobs. The answers Fig. 2: System Design for Personality Prediction System
chosen by the candidate in the personality
questionnairereveal type of the personality the person is The following are the research objectives of the work
having. Once the person knows the personality features of that focuses on Personality prediction.
him, he can choose the carrier optionsbest suitable for him.  The objective is classifying personalities and to analyse
Ensemble classifier is used to improve the test accuracy them based on the big five model with a given data set
score instead of a single classifier.Ensemble learning helps using classification algorithms and advanced data mining
to improvise machine learning predictions by combining concepts.
several models. This approach builds better predictive model  Using and exhibiting the data mining concepts and
compared to a single model. automate personality classification using python data
science libraries
 To design a system to improve performance of Multi Class
classification in Personality Prediction Analysis.

V. IMPLEMENTATION AND RESULT

Following are the four modules which are divided as

for every project it is a necessary to divide the collection of
source files required into individual functional unit. Every
module need to be built independently, tested and debugged.
 Data collection
 Attribute selection
 Pre-processing of data
 Prediction of personality

It is always necessary for any prediction system to

collect data and take decision about training dataset and test
dataset. Attributes in dataset are age, gender, answers to 40
questions in ‘yes’ and ‘no’, personality label based on
results. ‘elapse’ attribute removed as it is not contributing
towards classification.For better accuracy created the
ensemble model by combining support vector machine,
KNN, logistic regression, MLP and deep forest. The
accuracy of all the models alone was lesser ascompared to
the ensemble model.
Fig. 1: Algorithm implantation for APC

IJISRT22MAY1968 www.ijisrt.com 1601

Volume 7, Issue 5, May – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
The personalities are divided into 5 classes,
 Modest
 Semi Modest
 Confident
 Somewhat overconfident
 Overconfident

These classes were evaluated by combining the score

from all questions, and then quantizing it to the scale of 1 to
5
 For example, if the score is 10, then class will be 5*10/40
=1
 If score is 25, then class will be 5*25/40 = 3

Training was done for Deep Forest, k Nearest

Neighbor, Logistic Regression, Multilayer Perceptron and
Support Vector Classifiers.

Each classifier used the following configurations:

Fig. 4: Test F1, Test recall, Test Accuracy, Test precision
Deep Forest: Estimators = 100, Depth = 2
score for ensemble classifier
k Nearest Neighbor, Number of Neighbors =1
Logistic Regression, Solver=Limited-memory Broyden Classifier Test Accuracy Score
Fletcher–Goldfarb–Shanno Model, Total Iterations=1000 Deep Forest 0.598222222222222
Multilayer Perceptron, Number of hidden layers = 100
k Nearest Neighbour 0.736
Support Vector Classifier, Error Tolerance = 10-5
Logistic Regression 0.357333333333333
The classifiers were trained, and their classified Multilayer Perceptron 0.364444444444444
outputs were combined using Union Based Ensemble Support Vector 0.797333333333333
Learning Model. Ensemble classifier 0.931555555555555
Table 1: Classifiers with test accuracy score
This Model works via the following process,
 Combine all classes from all classifiers
 Identify unique classes 1
Deep Forest
 Check which classes are common between classifiers, and 0.9
discard them for better results 0.8
k Nearest
 Use remaining classes for evaluation of Precision, Recall, 0.7 Neighbour
fMeasure, and Confusion Matrix for the combined
classifiers. 0.6 Logistic
0.5 Regression
For any new input, classify it via all classifiers, and use 0.4 Multilayer
a Mode operation to get the final classification result to 0.3 Perceptron
identify different personality types 0.2 Support Vector

PySimpleGUI is used to create user interface. 0.1

Ensemble
0 classifier
Test Accuracy Score

Graph 1: Classifiers with test accuracy score

VI. CONCLUSION

Research in prediction and analysis of human being is

in great demand these days. Predicting the personality of the
candidatebythis system have made things easy in various
fields like recruitment process, medical counselling, and
likewise. Personality prediction using questionnairehelps to
find out the behaviorioulfeatures of the candidates taking the
survey. This paper focuses on providing a accurate system
forpersonality detection from questionnaire.

Fig. 3: GUI for personality detection

IJISRT22MAY1968 www.ijisrt.com 1602

Volume 7, Issue 5, May – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
REFERENCES [16.] “Rosenberg Self-Esteem Scale (SES) - Statistics
Solutions.” [Online]. Available:
[1.] Aleksandar Kartelj, Vladimir Filipović, Veljko http://www.statisticssolutions.com/
Milutinović, Novel approaches to automated rosenberg-self-esteem-scale-ses/. [Accessed:
personality classification: Ideas and theirpotentials. 28-Sep-2015].
[2.] Fazel Keshtkar, Candice Burkett, Haiying Li and [17.] Madarapu Soumica, Chamarthi Somasekhar Varma,
Arthur C. Graesser,Using Data Mining Techniques to Bobbili Siva Rama Krishna, 2021, Customer
Detect the Personality of Players in an Educational Personality Prediction using the Ensemble
Game Technique, INTERNATIONAL JOURNAL OF
[3.] G. Kumar and P. K. Bhatia, “Comparative Analysis ENGINEERING RESEARCH & TECHNOLOGY
of Software Engineering Models from Traditional to (IJERT) Volume 10, Issue 12 (December 2021),
Modern Methodologies,” in 2014 Fourth
International Conference on Advanced Computing &
Communication Technologies, 2014, pp. 189–196.
[4.] D. Leffingwell, Agile Software Requirements Lean
Requirements Practices for Teams, Programs, and the
Enterprise, 1st ed., MA: Addison Wesley, 2011.
[5.] “Manifesto for Agile Software
Development.”[Online].Available:
http://www.Agilemanifesto.org/. [Accessed:
20-Sep-2015].
[6.] “Principles behind the Agile Manifesto.”
[Online].Available:http://Agilemanifesto.org/principl
es.html. [Accessed: 20-Sep-2015].
[7.] Nurbiha A Shukora , Zaidatun Tasira, Henny Vander
Meijden(2015). An Examination of Online Learning
Effectiveness using Data Mining, Science Direct -
Procedia - Social and Behavioural Sciences 172, 555
– 562
[8.] D. Bishop and A. Deokar, “Toward an Understanding
of Preference for Agile Software Development
Methods from a Personality Theory Perspective,” in
2014 47th
[9.] L. Yan, Z. Mingyuan, and Y. Yongbo, “Risk
Correlation Analysis Based on Information
Management,” in 2010 3rd International Conference
on Information Management, Innovation
Management and Industrial Engineering, 2010, vol.
4, pp. 27.
[10.] H. Pretorius and H. Zaaiman, “Why use
communication training as enterprise-wide project
risk mitigation tool?,” Enterprise Systems
Conference (ES), 2013. pp. 1–6,.
[11.] “Personality and Values.”[Online]. Available:
http://saylordotorg.github.io/text_principles-of-mana
gement-v1.1/s06-0 2-personality-and-values.html.
[Accessed: 20-Sep-2015].
[12.] S. Zhu and L. Wang, “Research on software
undergraduates training countermeasures based on
the competency model,” in 2011 6th International
Conference on Computer Science & Education
(ICCSE), 2011, pp. 804–807.
[13.] T. Kanij, R. Merkel, and J. Grundy, “An empirical
study of the effects of personality on software
testing,” in 2013 26th International Conference on
Software Engineering Education and Training
(CSEE&T), 2013, pp. 239–248.
[14.] R. Kaplan and D. Saccuzzo, “Psychological Testing:
Principles, Applications, and Issues,” 2012, pp. 7–9.
[15.] S. John, O. P., & Srivastava, “Big Five Inventory
(BFI),” Handb. Personal. Theory Res., vol. 2, pp.
102–138, 1999.

IJISRT22MAY1968 www.ijisrt.com 1603