Detecting Fake Jobs

My third Metis project.

Goal

Detect fake job postings using a classification model.

Methods

• Oversample minority class (fraudulent job listings)
• Apply CountVectorizer to combined text fields
• Apply classification models
  • Logistic Regression
  • Random Forest
  • Naive Bayes
  • XGBoost

Results

Logistic Regression performed best on the validation data, detecting 78% of fraudulent listings, with a false postive rate of 20%.