Personality-Classification

This is a simple Logistic Regression for classifying the Big Five personality label from customer's review text.

Data Statistics:

There are 990 users in this dataset. Each user contains at least 350 words review text. Each user contains a Big Five personality scores.

Label of each personality class:

0: 'agreeableness'
1:'conscientiousness'
2: 'neuroticism'
3: 'extraversion'
4: 'openness'

Number of each personality class:

0: 628
1: 57
2: 40
3: 113
4: 52

Data Preprocessing:

Word2Vec 300d.

Traning/Testing Data Split

We split 890 for training set, 100 for testing set. We extract 20 samples from each classifier to make sure the testing set is balanced.

Upsampling classifier

Due to the imbalanced distribution of each persoanlity, we upsampled the 4 minority personality classes to '628'.