User Personality Prediction on Facebook Social Media using Machine Learning

International Journal of Trend in Scientific Research and Development (IJTSRD)
Volume 4 Issue 6, September-October 2020 Available Online: www.ijtsrd.com e-ISSN: 2456 – 6470
@ IJTSRD | Unique Paper ID – IJTSRD33414 | Volume – 4 | Issue – 6 | September-October 2020 Page 503
User Personality Prediction on Facebook
Social Media using Machine Learning
Poonam L Patil1, Dr. S. R. Jadhao2
2Assistant Professor,
1,2Department of Computer Engineering, R.H. Sapat College of Engineering,
Management Studies and Research Savitribai Phule Pune University, Nashik, Maharashtra, India
ABSTRACT
In recent years, Social network use is increasingly build-up. The various
statisticsare split widely through social media SuchasFacebook,Twitter.Data
about the person and what they communicate through the status updates are
important for researchin human personality. Thispaper intends to scrutinize
the forecasting of personality traits of Facebook users bases on machine
learning and part of the Big five model this experiment uses my personality
data set of Facebook users are used for linguistic factors respective to
personality correlation. We used the Data Prepossessing concept of data
mining after that feature Extraction. Next, we will work on feature selection.
The Personality Prediction system built in the XGboostingclassificationmodel.
KEYWORDS: social media, big five model, machine learning, personality
prediction, feature analysis, social network
How to cite this paper: Poonam L Patil |
Dr. S. R. Jadhao "User Personality
Prediction on Facebook Social Media
using Machine
Learning" Published
in International
Journal of Trend in
Scientific Research
and Development
(ijtsrd), ISSN: 2456-
6470, Volume-4 |
Issue-6, October 2020, pp.503-509, URL:
www.ijtsrd.com/papers/ijtsrd33414.pdf
Copyright © 2020 by author(s) and
International Journal of TrendinScientific
Research and Development Journal. This
is an Open Access article distributed
under the terms of
the Creative
Commons Attribution
License (CC BY 4.0)
(http://creativecommons.org/licenses/by/4.0)
INTRODUCTION
Now a Daysfrom social media like Facebook, Twitter,Reddit
Have become most trendy The propagation’s ofinternetand
intelligence technology, exclusively theonlinesocialnetwork
have revitalized how users communicate with other
electronically, the social media applicationsuchasFacebook,
Twitter, Instagram, Reddit not only introduce the written
and multimedia contain but also grant to circulate their
feelings, Moods, emotions online [1]. The Fig 1.shows that
the number of monthly social media users in India from Jan
2020- June 2020. Personality is the characteristic thewayof
thinking, feeling behaving. The distinct personality is
associated to the structure of various social relationsandco-
operations behavior on status profiles.
Our research predicts the personality based on user’s social
behavior and their language used for posting the status on
social media platformalthough the Facebookisthepresently
longer used to share photos, Video status. This accepts used
to predicate personality there for the goal of this research is
to build the prediction system thatcanautomaticallypredict
the user personality based on their activity on the social
Media [2].
A personality prediction model based on texts extracted
from social media that can be useful in several areas,
including marketing intelligence and social psychology, due
to the high volume of information generated and the
exposure level. The recognition of personality traits helpsto
find out the mutual conduct and may provide a subjective
view to text mining in social media, such as: sentiment
analysis, text clustering, and recommendation systems.
Figure. 1 number of social media users in India from 1
January 2020 to 20 June 2020 (using google trends)
First, we select the most favorable feature for each
personality major and portable forecast person personality.
Next, we proposed XGboosting method for predicting the
personality of social media user. we propose themethodone
category of social network Feature, we analyzed the
interrelationship between each of the personality traits.. In
the social network Feature some classes of anatomical
IJTSRD33414

International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
network properties such number of friends on social media
as well as their connections with personality traits. We
investigate the feature with larger co-relation with
Personality Traits. Finally, we proposed machine learning
algorithm for predication of personality by using Boosting.
The data collected by means of social media platform my
personality project data set.
A. Big Five Mode
Big five model is mostly used recently to measure the
personality. It describes the human personality structure. It
diminishes the greater number of personal objectives five
most personality traits. That model the composition OCEAN
[12][13]. The Table 1 Shows the five facts defining each of
factor with their characteristics [14] such as Openness,
Conscientiousness, Extraversion, Agreeableness,
Neuroticism.
 Openness: It is a general gratefulness for art, emotions,
imagination and variation of knowledge.
 Conscientiousness: It is tendency to display self-
discipline, try for achievements. The average level of
conscientiousness moves up among the young to adults
and then dismiss among the older to adults.
 Extra-version: It is distinguishing by spread of
activities, energy creation from external means.
 Agreeableness: It is Trait reflects the personality in
which the people are more cooperative. It reflects the
help-fullness personality.
 Neuroticism: It is negative traits of personality express
the sad emotions like depression, or moody.
TABLE1. OVERVIEW OF BIG FIVE MODEL
Personality Traits Some characteristics
Openness(O)
Openness measures the curiosity, tolerance, imagination, creativity, tolerance,
political liberalism, and appreciation for culture. People scoring high on Openness
such as switch, new and unusual ideas, appreciate and have a good sense of aesthetics
Conscientiousness(C)
Conscientiousness measures desire for a coordinated approach to life in comparison
to a spontaneous one. People scoring high on Conscientiousness are more likely to be
well organized, reliable, and consistent. people scoring high on conscientiousness
pursue long-term goals, enjoy planning and seek achievements.
Extraversion(E)
Extroversion measures a tendency to seek stimulation in the external world,to
express the positive emotions and give the company of others. People scoring high on
Extroversion tend to be more outgoing, friendly, and socially active. They are usually
energetic and talkative.
Agreeableness(A)
Agreeableness relates to a focus on maintaining positive social relations, being
friendly, compassionate, and cooperative. People scoring high on Agreeableness tend
to trust others and adapt to their needs. They also related to the short bound by social
trust and conventions
Neuroticism(N)
Neuroticism measures the tendency to experience mood swings andemotions such as
guilt, anger, anxiety, and depression.
The remaining of this paper is arrange as follows. We discussed the literature survey related to personality predication. we
describe the System Architecture and System Overview. we describe the System Analysis. Next, Result and Discussion.Finally,
we conclude our paper.
REVIEW OF LITERATURE
There are number of research paper on personality predication on social media Personality predication subjects divided in to
two methods: Computational Fundamentals Social network analysis.
Tandera et al. [2]They used the two datasets, one from mypersonaliy Facebook dataset and other is manuallycomposed.They
Predict the Personality of the person by using the BigFiveModel. Using the Support vector machine,theyachievedthetopmost
Prediction accuracy of 70.40%.
Pennebaker Key et al. [3] Introduce work related to personality extraction from the text they examine the words in different
factors such as diaries, college assignments and social psychological manuscripts to observe the personality related features
with linguistic library. The Result shows that Agreeableness personality trades tents to use more text the neuroticism used
more negative/sad words.
Ana CES lim et al. [4]wrote a pioneering work dedicated to personality prediction into a multi-label classification problem. In
that, they process more than one personality trait. They were classified the personality traits of Twitter using Naïve Bayesian
prediction model.
Argaman et al [5] They classified the personality traits namely neuroticism and Extraversion using Lingui sting feature. They
observed that neuroticism iscorrespondence to the functional lexical feature and the extraversion trait resultinlessobserved.
In N.M.A listeria et al [6] They introduced the Naïve Bayesian method for the classification of personality traits. In that, the
Naive Bayesian method consists of two phases such as the Learning phase and classification phase. The user- written text is
used as input for predicting the personality then match them to find the partner on online dating sites.
SoujanyaPoria et al. [7] They proposed a new approach for personality detection which is based on incorporating the
sentiment, affective and common-sense knowledge from the text using resources. In their approach, they combined common

sense knowledge-based features with phycho-linguistic features and frequency-based features and later the features were
employed in supervised classifiers. Further, they developed five support vector machinemodelsforfivepersonalitytraits.They
designed five Social Media Optimization (SMO) based supervised classifiers for five personality traits. Their experimental
results show that the use of common-sense knowledge with perceptual and sentiment information with psycho-linguistic
features and frequency-based analysis at lexical level that upgrades the accuracy of the current frameworks.
Go beak et al [8] predicted the personality of 279 Facebook users. In which the find the word count as Linguistic feature and
friend count as SNA feature.
Sibel Adult et al [9] Predict the personality of user from Facebook Data and text from Twitter. They introduced the number of
measures related to number of social media. They analyzed these features based on textual analysis of message send by
another user. The aim of our study examines the all personality traits from the structure of social network analysis to the
personality interaction using my personality project dataset [10] as well as Facebook API.
Ong e al. [11] Predict the personality based on Twitter information in Bahasa Indonesia. The system uses the 329 users of
Twitter social media to predict the personality. They use the XGboost classification model.
SYSTEM OVERWIEW
A. Problem statement
To effectively evaluate the performance of XGBoost usingmachine learning for the accuracyofpersonalitypredictionofuseron
social media.
B. system Architecture
Personality Prediction model consist of following terms
1. Data Preprocessing
All the data goes through preprocessing stages before it processed. Preprocessing perform steps. Consist of removing URLs,
Symbols, Names, Stemming, removing stop word and lower cases. In our work we use python and machine learning. Data
before using the machine learning perform the data preprocessing.
2. Feature Extraction
A User’s behavior on social media isoffered by current behavior ofanother user. In many applicationsavailable for explaining
such behavior happen and expand [15]. In our work all data from Dataset classified in two parts:
 Text Feature Extraction: Analyze the content of social media status texts uses the dictionaries.
 Social network behavior analysis:Which consist ofNetwork size, Transitivity, Brokerage,Densitythisinformationdenotes
the users network behavior on Facebook.
Figure 2 System overview
Before the text data feed in machine learning for extorting the feature, the raw text status represented in the following forms:
Bag ofwords representation [16]: In this presentation every sentence is the different set of words in whichwedon’tconsidera
grammar here repetition of word together in feature future classification.
SNA (Social Network Analysis): Itis method ofcollectingand examining data from social network suchasFacebook,tweeter
and Instagram. In our study we used feature related to social network of user with personality trades such as network size
betweenness, density, brokerage and Transitivity.

Network size: This introduce the number of social friends on social media [17].
Betweenness: Refers the number of parks between pair of individuals those are not connected to each other directlythrough
the density it indicates the potential connection on network that are actual connections more density network induces more
dissipation between persons in information flow [18].
Brokerage: It is a state or situation in which person connects another people and the unconnected action or fills the gap in
social structure.
Transitivity: it is based on friend of my friend is also my friend in which single are directly connected to each other one of
them is only accessible with other individual represent them frequency of interaction between the network nodes [19] [20].
3. Feature Selection
For constructing classification model the feature connection is important. feature selection is preprocessingstepforamachine
learning. feature selection is use for dimensionality reduction and can effectively find the both irrelevant and redundant
features. There are two methods to major the correlation between two random variables.
Based on Classified Liner correlation.
Based on information theory [21].
To Measure the stability of linear correlation between two variable and criticize the important features for personality traits
prediction we use the Pearson Correlation coefficient. It isa part of the linear correlation between two variables X and Y [21].
For the pair of variable(X,Y),the linear Correlation coefficient formula of r(X,Y) is given by:
(3)
Where
(4)
And
(5)
These are the mean of the X and Y Variable and n is sample size. The value of correlation coefficient in between -1 and 1. If X
and Y Variable arecompletely parallel then r(X, Y) takes the Value 1 as Positive Correlation or for NegativeCorrelationittakes
the value as -1. If both variables are independent on each other it takes the value as Zero [22][23].
4. Prediction Model
After finding the correlation between the features, we apply the classifier for predicting the personality traits.
We use the XGBoost model as classifier. XGBoost is an algorithm. Also, it has recently been dominating applied machine
learning. XGBoost is a gradient boosted decision trees implementation. It is a type of software library in Python. There are
number of interfacesavailable to access thismodel suchasPython interfacealong with integrated modelinscikit-learn.weuse
the Python interfaces for XGBoost model. The Algorithm for Building the XGBoosting Model Perform the following Steps:
 Load All the Python Libraries Here We load all the libraries of python such as XGBoost, Readr, Stringr.
 Next part is to load all collected Dataset (Here We Use the mypersonality Dataset of Facebook User)
 First Load the label Of Train Data.
 Next Combine the Training and Testing Data.
 Perform Data Cleaning
 Here All the Feature are Categorized in various format and Perform the Data Prepossessing
 Splitting of Training and Testing Data.
 Tune and Run he Model.
 Predicting score test set.
XGboosting algorithm working
1. First model F0 is defined to predict the target variable Y.
2. fit model to the residual h1(x) =Y-F0.
3. Create a new model using the h1(x) and F0 to give the F1, it is boosted version of F0.
4. The mean squared error of F1 will be lower than the F0.
5. To Improve the performance of F1 model, then residuals of F1 and create a new model F2. F2(x)<-F1(x)+h2(x)
6. This can be done for the ‘m’ iterations.

RESULT AND ANALYSIS
Table 2 shows that the word “love” shows the positive emotions to words the extraverted personality traits. Table shows the
Probability score of all personality traits and category of the personality. Table 3 shows the personality score of the “python”
word. The Python word gives the openness type of personality trait. Based on the correlation results for the social network
features, we found that extraversion represents the highest correlated trait.
TABLE 2 SHOWS THE RESULT OF THE TEXT ANALYSIS OF THE “LOVE” WORD
Text analysis Personality traits Prediction Score Prediction Probability score Trait Category
“Love”
Openness(O) 0.8692 4.319 True
Conscientiousness(C) 0.4149 3.662 False
Extraversion(E) 0.4452 3.472 True
Agreeableness(A) 0.5590 3.776 True
Neuroticism(N) 0.5982 2.627 True
TABLE 3 SHOWS THE RESULT OF THE TEXT ANALYSIS OF THE “PYTHON” WORD
Text analysis Personality traits Prediction Score Prediction Probability score Trait Category
“Python”
Openness(O) 0.664 4.123 True
Conscientiousness(C) 0.441 3.518 False
Extraversion(E) 0.351 3.330 False
Agreeableness(A) 0.424 3.393 False
Neuroticism(N) 0.461 2.808 False
Table 4 shows the personality analyzer whichindicate that the all profiles are collected from the social media andcalculatethe
personality score.
TABLE 4 SHOWS THE RESULT OF THE SOCIAL MEDIA FRIENDS PERSONALITY PREDICTION SCORE
Users
Prediction Score Prediction Probability Score
O C E A N O C E A N
User1 0.98 0.19 0.21 0.67 0.17 4.23 3.14 3.29 3.49 2.71
User2 0.64 0.71 0.23 0.6 0.50 3.67 3.49 3.23 3.68 3.01
User3 0.99 0.10 0.14 0.12 0.05 4.20 3.19 3.38 3.04 2.53
User4 0.71 0.39 0.37 0.51 0.38 4.16 3.43 3.37 3.49 2.70
User5 0.66 0.44 0.35 0.42 0.47 4.12 3.51 3.21 3.39 2.82
The Personality Prediction score of the all five personality traitsare shown in the figure 3. The result shows that the openness
personality traits give the highest personality score after prediction.
Figure 3 Shows the personality score of all Five personality traits prediction
We compare the result of accuracywith the existingmachine learning algorithm. Figure 4showsthatthepersonalityprediction
result analysis. Finally, for achieving maximum prediction accuracy, Xgboosting model is used for the maximum accuracy.
TABLE 4 SHOWS THE ACCURACY OF ALL PERSONALITY TRAIT IN PROPOSED AND EXISTING SYSTEM
Performance Measures Personality Traits Proposed System % accuracy Existing system % accuracy
Accuracy
Openness(O) 78.66 68
Conscientiousness(C) 64.5 61.6
Extraversion(E) 76 44
Agreeableness(A) 60.8 54.8
Neuroticism(N) 64 64.8

Figure 4 Performance analysis
CONCLUSION
Social network analysis hasincreased largelyinrecenttimes.
To extract the personality of any person on the social
networking websites is very useful for many applications in
various domains like including job success, attractiveness,
and happiness.Personality detection from social media is to
extract the feature from there updates and the behavior
attribute of a person from the written text on social media.
This Prediction Model help to predict the personality of user
from social media. Xgboosting predictionmodeloutperforms
than the other prediction models on the all personalitytraits.
Acknowledgment
I have huge delight in presenting the project” User
Personality prediction on Social Media Using Machine
Learning” under the guidance of Dr. S. R. Jadhao and PG
Coordinator Prof. A. S. Vaidya. I gratefully thanks our HOD
Dr. D. V. Patil to Gokhale Education Society’s R. H. Sapat
College of Engineering, Management Studies Research,
Nashik-5 for giving required equipment’s, web get and
Research Books. I also Thank to Teaching and Non-teaching
staff of GESRHSCOE who helped me.
References
[1] Islam Md. Rafiqul, Kabir Ashad, Ulhaq Anwar, wang
Hua,, “Depression Detection from Social network data
using Machine Learning techniques,” by springer
Nature Switzerland AG 2018.
[2] Tandera T, Hendro, Suhartono D, Wongso R, Yen Lina
Praseti, “Personality prediction system fromFacebook
users,” Procedia Comput. Sci., vol. 116, pp. 604–611,
Dec. 2017.
[3] J. W. Pennebaker, R. L. Boyd, K. Jordan, and K.
Blackburn, “The development and psychometric
properties of LIWC2015,” Tech. Rep. 2015.
[4] Lima, Ana CES, and Leandro N. De Castro, “Multi-label
Semi-supervised Classification Applied to Personality
Prediction in Tweets,”Computational Intelligence and
11th BrazilianCongress on Computational Intelligence
(BRICS-CCI and CBIC), 2013 BRICS Congress on. IEEE,
2013.
[5] S. Argamon, S. Dhawle, M. Koppel, and J. Pennebaker,
“Lexical predictors of personality type,” Tech. Rep.,
2005.
[6] N. M. A Lestari, I. K. G. D .Putra, A. A. K. A. Cahyawan,
“Personality types classification for Indonesian text I
partners searching website using Na¨ıve Bayes
Methods”, International Journal of software and
Informatics Issue,2013.
[7] SoujanyaPoria, AlexandarGelbukh, Basant Agarwal,
Erik Cambria, and Newton Howard, “Common Sense
Knowledge Based Personality. Recognition fromText”,
Springer-Verlag Berlin Heidelberg pp. 484–496, 2013.
[8] J. Golbeck, C. Robles, and K. Turner, “Predicting
personality with social media,”in Proc.ExtendedAbstr.
Hum. Factors Comput. Syst. (CHI), 2011, pp. 253–262.
[9] S. Adali and J. Golbeck, “Predicting personality with
social behavior,” in Proc. IEEE/ACM Int. Conf. Adv.
Social Netw. Anal. Mining (ASONAM), Aug. 2012, pp.
302–309.
[10] M. Kosinski, S. C. Matz, S. D. Gosling, V. Popov, and D.
Stillwell, “Facebook as a research tool for the social
sciences: Opportunities, challenges, ethical
considerations, and practical guidelines,” Amer.
Psychol., vol. 70, no. 6, pp. 543–556, 2015.
[11] V. Ong; et al.,” Personality prediction based on twitter
information in Bahasa Indonesia,” in Proc. Federated
Conf. Comput. Sci. Inf. Syst. (FedCSIS), 2017, pp. 367–
372.
[12] L. R. Goldberg, “The structure of phenotypic
personality traits,”Amer.Psychol., vol. 48, no.1,pp.26–
34, 1993.
[13] E. C. Tupes and R. E. Christal, “Recurrent personality
factors based on trait ratings,”J. Pers., vol. 60, no. 2, pp.
225–251, 1992.
[14] O. P. John and S. Srivastava, “The big five trait
taxonomy: History, measurement, and theoretical
perspectives,” Handbook of Personality: Theory and
Research, vol. 2. 1999, pp. 102–138
[15] T. P. Michalak, T. Rahwan, and M. Wooldridge,
“Strategic social network analysis,” inProc.AAAI,2017,
pp. 4841–4845.
[16] Soumya George K,Shibily Joseph, “TextClassificationby
Augmenting BagOf Words(BOW) Representationwith
Co-occurencefeaure”, IOSR Journal of computer
Engineering, 2014, pp 34-38.

[17] J. E. Lonnqvist, J. V. Itkonen, M. Verkasalo, and P.
Poutvaara, “The ﬁve factor model of personality and
degree and transitivity ofFacebook social networks,”J.
Res. Person., vol. 50, pp. 98–101, Jun. 2014
[18] H. Lin and L. Qiu, “Sharing emotion on Facebook:
Network size, density, and individual motivation,” in
Proc. Extended Abstr. Hum. Factors Comput. Syst.
(CHI), 2012, pp. 2573–2578.
[19] M. E. J. Newman and J. Park, “Why social networks are
different from other types of networks,” Phys. Rev. E,
Stat. Phys. Plasmas Fluids Relat. Interdiscip. Top., vol.
68, no. 3, p. 036122, 2003.
[20] M. Aghagolzadeh, I. Barjasteh, and H. Radha,
“Transitivity matrix ofsocial network graphs,” in Proc.
IEEE Stat. Signal Process. Workshop (SSP), Aug. 2012,
pp. 145–148.
[21] L. Yu and H. Liu, “Feature selection for high-
dimensional data: A fast correlation-based ﬁlter
solution,” in Proc. 20th Int. Conf. Mach. Learn. (ICML),
2003, pp. 856–863.
[22] Cramer, Fundamental, “Statistics for Social Research:
Step-by-Step Calculations and Computer Techniques
Using SPSS for Windows,” Evanston, IL, USA:
Routledge, 2003.
[23] Hall M A, “Correlation-Based Feature Selection for
Machine Learning,” 1999.

User Personality Prediction on Facebook Social Media using Machine Learning

More Related Content

Similar to User Personality Prediction on Facebook Social Media using Machine Learning (20)

More from ijtsrd (20)

Recently uploaded (20)

User Personality Prediction on Facebook Social Media using Machine Learning