Big Five Personality Test <!-- omit in toc -->

Table of contents <!-- omit in toc -->

• Pipeline
• Data
• Modelling
• References The following project developed a solution to grade Big Five Personality Tests. The project was developed according to CRISP-DM phases, as follows:

<center><img src="assets/CRISPDM.png" alt="CRISPDM" style="width: 500px;"/></center> <center><i>Figure 1. CRISP-DM Methodology.<br>Source: [1-2].</i></center>

The target variable corresponds to the trait with max value of the test. Further information about the traits can be found below.

| Trait | Class | Description | | :--------------------: | :---: | :------------------------------------------------------------------------------------------------------------------------------------- | | Extroversion | E | High scores: People who tend to be social<br>Low scores: People who prefer to work alone in projects | | Agreeableness | A | High scores: Politically correct people<br>Low Scores: Direct people | | Conscientiousness | C | High Scores: People who tend to follow the rules and prefer order<br>Low Scores: People who tend to be disorganized | | Neuroticism | N | How emotional the person can be | | Openness to Experience | O | High Scores: People who tend to “dream with their eyes open”<br>Low Scores: People who tend to “have their feet on the ground” |

Pipeline

The following pipeline summarizes the steps and procedures which have been taken into account during training / validation and testing.

<center><img src="assets/Pipeline.png" alt="Pipeline" style="width: 900px;"/></center> <center><i>Figure 2. Pipeline.<br></i></center>

Data

The exploratory Data Analysis allowed us to conclude that there were a imbalance in target variable. Then two approaches to go over this situation were followed:

• UnderSampling
• Oversampling

Data can be downloaded here.

Modelling

The project at hand corresponds to a problem of multiclass classification. The following classifiers were trained and validated:

• Logistic Regression Classifier
• Decision Tree Classifier
• Random Forest Classifier
• Naive Bayes Classifier

References

[1]	M. F. Hornick, E. Marcadé, and S. Venkayala, “Chapter 3 - Data Mining Process,” in The Morgan Kaufmann Series in Data Management Systems, M. F. Hornick, E. Marcadé, and S. B. T.-J. D. M. Venkayala, Eds. Burlington: Morgan Kaufmann, 2007, pp. 51–83.
[2]	“File:CRISP-DM Process Diagram.png - Wikimedia Commons.” .
[3]	B. Tunguz, L. Petar, and M. Akdag, “Big Five Personality Test | 1M Answers to 50 personality items, and technical information.” [Online]. Available: https://www.kaggle.com/tunguz/big-five-personality-test. [Accessed: 20-Jun-2020].