Income Prediction Project

Overview

Binary classification project to predict whether annual income is greater than $50K.

• Target: income
• Metric: (F1 Score + AUC) / 2

Dataset

• train.csv: 39,073 rows × 15 columns
• test.csv: 9,769 rows × 14 columns
• Test set does not include income.

Current Progress

Preprocessing

• Loaded train/test datasets
• Checked column types and missing values
• Found missing values in categorical columns:
  • workclass
  • occupation
  • native_country
• Filled categorical missing values using train-set mode as baseline
• Converted target label:
  • <=50K → 0
  • >50K → 1
• Applied one-hot encoding to categorical features

Output

Current preprocessing creates:

• X_train: (39073, 105)
• X_test: (9769, 105)
• y_train: encoded target labels
• test_id: id column for submission

Next Steps

• Confirm preprocessing strategy with team
• Train baseline models
• Evaluate using F1 and AUC
• Generate prediction.csv