Classification/preprocessing/ishwari

This MR includes all preprocessing steps for the census income dataset, including:

1. Cleaned missing values (?, NaN, Null, Na etc.) and standardized text.
2. Dropped irrelevant or skewed features (capital-gain, capital-loss, education-num).
3. Performed EDA by demonstrating distribution of numerical and categorical features, heatmap, correlation matrix, boxplot for outliers.
4. Encoded categorical features using appropriate strategies:
   • Binary encoding for sex
   • One-hot encoding for race, relationship, marital-status, workclass, and grouped native-country
   • Grouped rare/unknown categories to reduce sparsity
5. Cleaned numerical features by clipping extreme outliers in hours-per-week.

Finalized processed dataset and notebook

• Dataset saved under: data/processed/
• Notebook: notebooks/classification/preprocessing_ishwari.ipynb