Visualization

This section covers visualization tools for analysis and interpretation.

Overview

NIRS4ALL provides comprehensive visualization tools for analyzing predictions, understanding model behavior, and interpreting results.

� In-Pipeline Charts

Visualize spectra, folds, targets, augmentation, and exclusions during pipeline execution.

Pipeline

In-Pipeline Charts
📊 Prediction Charts

Visualize predictions, residuals, and model performance.

Analysis

Analyzer Charts Reference
🔀 Pipeline Diagram

Visualize pipeline structure as interactive diagrams.

Structure

Pipeline Diagram
🔍 SHAP Analysis

Explain model predictions with SHAP values.

Explainability

SHAP Analysis for NIRS Models

Quick Example

import nirs4all
from nirs4all.visualization import PredictionAnalyzer

# Run pipeline and get results
result = nirs4all.run(pipeline, dataset="data/")

# Create analyzer
analyzer = PredictionAnalyzer(result)

# Generate charts
analyzer.plot_predictions()      # Predicted vs actual
analyzer.plot_residuals()        # Residual analysis
analyzer.plot_calibration()      # Calibration curve

Available Charts

Chart Type

Description

Use Case

Predictions

Scatter plot of predicted vs actual

Model accuracy

Residuals

Residual distribution and patterns

Bias detection

Calibration

Reliability diagram

Probability calibration

Learning Curves

Performance vs training size

Data sufficiency

See Also