easystats/performance

Features

• R - Provides standardized methods to assess model quality, predictive accuracy, and statistical assumptions within the R programming language.
• Model Agnostic Diagnostics - Decouples diagnostic logic from specific statistical frameworks to enable uniform validation of assumptions across frequentist and Bayesian models.
• Bayesian Model Comparison - Ranks and compares multiple statistical models using performance metrics and information criteria to select the best fit for data.
• Model Performance Metrics - Computes standard performance indices such as RMSE, AIC, and BIC to assess predictive accuracy and reliability.
• Comparative Analyzers - Calculates and visualizes performance indices across several models simultaneously to determine the best fit for data.
• Statistical Comparison Frameworks - Assesses differences between multiple models using likelihood ratio tests and Bayes factors to determine relative quality.
• Statistical Model Validators - Checks statistical assumptions and identifies potential issues like multicollinearity or overdispersion to ensure models are reliable and accurate.
• Goodness of Fit Metrics - Calculates goodness-of-fit indices, diagnostic metrics, and performance scores for frequentist and Bayesian statistical models.
• Statistical Diagnostics - Generates automated visualizations and metrics to detect outliers, residual patterns, and structural failures in various statistical models.
• Custom Accuracy Metrics - Quantifies prediction accuracy using standard error metrics, information criteria, and classification scores.
• Multicollinearity Analyses - Quantifies multicollinearity within statistical models to identify potential issues with predictor variables.
• Model Comparison Tools - Provides a set of tools for ranking and comparing multiple statistical models using information criteria, Bayes factors, and normalized performance indices.
• Model Diagnostics - Identifies statistical violations such as overdispersion, zero-inflation, heteroskedasticity, and singular fits to ensure model assumptions are met.
• Diagnostic Visualizers - Generates graphical representations of model assumptions and performance metrics to facilitate the identification of patterns and structural failures.
• Classification Evaluators - Plots receiver operating characteristic curves to compare the diagnostic ability and predictive accuracy of classification models.
• Determination Coefficient Calculators - Computes coefficients of determination to measure the proportion of variance explained by a model.
• Performance Rankers - Generates a normalized performance score for each model to provide a clear heuristic ranking of overall model quality.
• Performance Summarizers - Computes comprehensive quality metrics and fit indices for diverse statistical models to provide a clear overview of performance.
• Performance Comparison Visualizers - Creates spiderweb plots of normalized performance indices to visually compare the relative strengths and weaknesses of multiple models.
• Model Weighting Utilities - Computes relative model weights based on information criteria to quantify the evidence supporting each individual model.
• Statistical Residual Analyzers - Visualizes binned residuals to identify patterns or deviations that indicate whether a statistical model accurately represents the underlying distribution of the observed data.
• Lazy Evaluation - Defers heavy statistical computations until requested to optimize performance when analyzing large sets of candidate models.
• S3 Dispatchers - Uses language-specific method dispatch to apply consistent statistical calculations across diverse model objects regardless of their underlying implementation.
• Predictive Accuracy Validators - Compares observed data against simulated data from the model to verify if the model successfully captures the variation of the original dataset.
• Psychometric Reliability Analysis - Provides tools for assessing the internal consistency and reliability of psychometric scales and questionnaire items.
• Psychometric Reliability Analyzers - Evaluates the internal consistency and quality of questionnaire items and scales using standard reliability coefficients and correlation analysis.
• Hypothesis Testing - Executes hypothesis tests such as Bayes factors and nested model comparisons to determine statistical significance.
• Metric Normalizers - Transforms disparate statistical outputs into standardized scales to allow direct comparison and ranking of heterogeneous model types.
• Statistical Assumption Validators - Performs diagnostic tests and visual checks to verify statistical assumptions such as normality, homoscedasticity, independence of residuals, and the absence of influential outliers.

Star history