Preprocessing

The preprocessing module in sorix offers a comprehensive set of tools for data cleaning and feature engineering. These utilities transform raw data into a numerical format suitable for deep learning models, ensuring numerical stability and faster convergence.

Feature Scaling

• Scalers: Normalize or standardize numeric features. Includes MinMaxScaler, StandardScaler, and RobustScaler.

Categorical Encoding

• Encoders: Transform categorical variables into numerical representations. Currently supports OneHotEncoder.

Pipeline Integration

• ColumnTransformer: Apply different transformations to different columns in a single, reusable object.