#TITLE_ALTERNATIVE#

• Main Author: Wijaya (NIM: 18214028), Santo
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/30810
• Institution: Institut Teknologi Bandung

<p align="justify">Nowadays, Social media become a life style in society, especially in Indonesia. Social media user tend to express their feeling and idea through social media. Thus, there is possibility to use social media data for encourage psychologycal assessment. There are a lot of data mining techniques developed today. Furthermore, There are a lot of psychologycal theory developed by researchers.. This research focused on comparing data mining technique and psychologycal theory that appropriate to be used for social media data. Support Vector Machine and Naive Bayes will be compared to identify which has a good performance to classify social media data. Psychological Theory, MBTI and Five Factor Model will be compared using prediction accuracy from the machine. Variation of the technique is using kernel (for SVM), tokenizing (Unigram, Bigram), and weighting (TF,TF-IDF). Based on the result, Support Vector Machine (using linear kernel, Unigram+Bigram, TF-IDF) is better tehcnique than Naive Bayes with 74,19% of accuracy. Moreover, MBTI theory is more appropiate to identify psychological assessment using social media data than Five Factor Model because of Its accuracy that reach 66,7% in first scheme and 62,5% in scheme 2.<p align="justify">