A prediction model to determine if a person's income is over $50,000 a year.
The dataset is extracted from the 1994 Census database and is available on the UCI repository. The size of the dataset is 48,842 rows and includes 14 attributes such as age, gender, occupation, number of hours the individual works per week, etc.
- Exploratory data analysis (uni-variate and bi-variate)
- Data preprocessing (deduplication, handling missing values)
- Classification using logistic regression
- Classification using a gradient boosting machine
- Feature engineering
| Algorithm | Accuracy | Area under the curve |
|---|---|---|
| Logistic regression | 81.63% | 0.862 |
| Gradient boosting machine | 82.58% | 0.881 |