DistrictDataLabs/yellowbrick
Yellowbrick
Yellowbrick is a machine learning visualization library and model diagnostic tool designed to analyze feature importance, target distributions, and model error metrics. It serves as a visual toolkit for diagnosing underfitting and overfitting through the use of validation and learning curves.
The project provides specialized suites for evaluating predictive models and unsupervised learning. It enables the determination of optimal cluster counts via elbow methods and silhouette coefficients, and assesses classifier and regressor quality through ROC curves, confusion matrices, and residual plots.
The library covers several high-level capability areas, including feature engineering analysis to identify predictive variables, hyperparameter tuning to adjust model complexity, and regression error diagnosis to identify influential data points. It also includes tools for manifold learning projection to visualize high-dimensional data and text corpora.
The tool integrates with the Scikit-Learn API to consume standard fit and predict methods.
Features
- Model Diagnostic Tools - Provides a comprehensive suite of visual diagnostics to evaluate model performance and tune hyperparameters.
- Visual Model Selection Tools - Provides visual diagnostic tools to compare different algorithms and choose the best model for a specific dataset.
- Performance Visualizations - Provides graphical tools for analyzing classification model performance, including precision-recall plots and confusion matrices.
- Feature Importance Ranking - Identifies the most predictive variables through importance ranking, correlation matrices, and recursive elimination.
- Feature Importance Visualization - Uses ranked diagrams to visualize the relationship between features and target metrics to identify influential predictors.
- Feature Selection Methods - Ranks feature importance and performs recursive elimination to identify the most predictive variables.
- Cluster Count Selection Methods - Implements elbow methods and silhouette coefficients to determine the optimal number of clusters.
- Regression Visualization Tools - Provides residual plots and Cook's distance visualizations to diagnose linear regression model breakdowns.
- Machine Learning Visualization Libraries - Provides a comprehensive set of plotting utilities for analyzing feature importance, target distributions, and model error metrics.
- Model Selection and Validation - Provides a visual toolkit for diagnosing underfitting and overfitting via validation and learning curves.
- Model Evaluation Suites - Offers a suite for generating ROC curves, confusion matrices, and residual plots to assess model quality.
- Visual Hyperparameter Search - Provides visual tools to analyze model performance across different hyperparameter settings to identify underfitting and overfitting.
- Hyperparameter Tuning - Generates validation and learning curves to systematically tune model hyperparameters.
- Model Performance Evaluators - Generates diagnostic plots like ROC curves to quantify accuracy and trade-offs in model performance.
- Model Performance Visualizations - Generates diagnostic plots for precision, recall, and error metrics to evaluate machine learning estimators.
- Error Diagnostics - Plots residuals and prediction errors to identify specific data instances causing model breakdowns.
- Precision-Recall Curve Generators - Provides ROC curves, precision-recall reports, and confusion matrices to evaluate classifier performance.
- Residual Analysis - Calculates residuals between predictions and actual values to generate diagnostic plots for identifying model outliers.
- API Consumers - Integrates with Scikit-Learn estimators by consuming standard fit and predict methods to extract metadata for plotting.
- Diagnostic Tools - Provides specialized tools for determining optimal cluster counts and evaluating unsupervised learning quality.
- Regression Diagnostics - Analyzes model breakdowns and influential data points through residual plots and Cook's distance.
- Clustering Evaluation Metrics - Implements silhouette coefficients to evaluate the quality and optimal count of unsupervised clusters.
- Feature Correlation Analysis - Produces correlation matrices and scatter plots to evaluate statistical relationships and identify collinearity between features.
- High-Dimensional Feature Visualization - Plots high-dimensional features using projection techniques and parallel coordinates to detect complex data patterns.
- Evaluation Visualizers - Renders visual diagnostics across feature analysis and model evaluation to illustrate the iterative ML building process.
- Influence Analysis - Uses residual and error plots to determine the influence of specific instances on linear regression results.
- Recursive Feature Elimination - Provides tools to iteratively remove unimportant predictors and identify the most influential data subset.
- Target Distribution Analysis - Visualizes class imbalances and feature correlations to understand how target variables relate to input data.
- Data Characteristic Analyzers - Generates histograms and correlation maps to examine attribute distributions and data patterns.
- Manifold Learning - Implements non-linear dimensionality reduction like t-SNE and UMAP to visualize high-dimensional data in 2D or 3D.
- Model Interpretation - Visual diagnostic tools for machine learning model selection.
- Observability And Monitoring - Visual diagnostic tools for model selection and analysis.
- Automated EDA and Visualization - Visual diagnostic tools for machine learning.
- Computation and Optimization - Visualization tools for machine learning model evaluation.
- Visualization - Visual diagnostic tools for machine learning models.
