ds-utils-preprocess

You can visualize feature distributions by invoking preprocessing functions on your DataFrame to inspect a single feature's distribution based on its data type. This helps understand data shape and identify outliers during model preparation.

You can visualize correlations by generating a correlation heatmap and then identifying highly correlated feature pairs. This reveals linear relationships between variables and helps reduce multicollinearity before modeling.

A correlation dendrogram reveals hierarchical relationships among features by clustering them based on correlation distance. This visualization helps identify groups of redundant variables for feature selection and dimensionality reduction.

You can measure feature dependency by computing mutual information between features and a target label. This quantifies how much information the feature provides about the target, supporting effective feature selection beyond linear correlations.

Yes, you can assess PCA explained variance and plot feature interactions directly. The skill operates with no external dependencies, allowing you to evaluate principal components and explore joint feature relationships seamlessly.

You can summarize statistics by category by extracting a statistics dataframe per label. This computes summary metrics for features grouped by their target category, providing detailed insights into class-specific data distributions.