Test 10: Matrix Features

Purpose

Visualize and analyze the distribution of important model features, helping to detect outliers, unusual values, or "junk" data that could affect model performance.

Required Inputs

From configs/env.sh:

• WORK_DIR: Output directory for results
• MODEL_PATH: Path to the model
• REPOSITORY_PATH: Path to the data repository
• TEST_SAMPLES: Path to the test samples
• config/feat_resolution.tsv: (Optional) Defines resolution for feature plots

Depends On

• Test 05: But Why: Uses feature importance to select which features to plot

How to Run

From your TestKit folder, execute:

./run.specific.sh 10

Or include as part of the full suite:

./run.sh

What This Test Does

• Generates matrix plots for selected features, highlighting their value distributions
• Uses feat_resolution.tsv (if provided) to control plot resolution
• Helps identify outliers, missing values, and distribution issues in processed features

Output Location

• ${WORK_DIR}/outputs/features_stats.tsv: Table with statistics for each feature (mean, std, missing value percentage, split by cases/controls)
• ${WORK_DIR}/outputs/graphs: Graphs for each feature showing value distributions

How to Interpret Results

• Review feature statistics for reasonable values and balance
• Inspect graphs for outliers, skewed distributions, or unexpected patterns
• Use findings to refine feature engineering and improve model robustness