graph LR
Regression_Modeling["Regression & Modeling"]
linear_regression["linear_regression"]
logistic_regression["logistic_regression"]
mediation_analysis["mediation_analysis"]
_point_estimate["_point_estimate"]
_relimp["_relimp"]
_pval_from_bootci["_pval_from_bootci"]
_bias_corrected_ci["_bias_corrected_ci"]
mediation_analysis -- "Calls" --> linear_regression
mediation_analysis -- "Calls" --> logistic_regression
mediation_analysis -- "Calls" --> _point_estimate
mediation_analysis -- "Calls" --> _pval_from_bootci
mediation_analysis -- "Calls" --> _bias_corrected_ci
_point_estimate -- "Calls" --> linear_regression
_point_estimate -- "Calls" --> logistic_regression
click Regression_Modeling href "https://github.com/CodeBoarding/GeneratedOnBoardings/blob/main/pingouin/Regression_Modeling.md" "Details"
The pingouin.regression subsystem provides a comprehensive suite for statistical regression analysis. The Regression & Modeling component serves as the primary entry point, orchestrating calls to specialized regression algorithms such as linear_regression and logistic_regression. For complex analyses like mediation_analysis, the system leverages these core regression models and integrates utility components like _point_estimate, _pval_from_bootci, and _bias_corrected_ci to derive robust statistical inferences. _point_estimate directly interacts with the regression models to extract necessary coefficients, while _pval_from_bootci and _bias_corrected_ci process bootstrap results to provide refined statistical measures. The _relimp component operates independently to assess predictor importance within models. This modular design ensures clear separation of concerns, with core regression models handling computations and utility components refining or interpreting their outputs.
Regression & Modeling [Expand]
The overarching component that provides the public API for all regression-related functionalities. It acts as the entry point for users to access linear, logistic, and mediation analyses, orchestrating the underlying helper functions and algorithms.
Related Classes/Methods:
Implements the core algorithm for fitting ordinary least squares (OLS) linear regression models. It computes coefficients, R-squared, p-values, and other relevant statistics for continuous outcome variables.
Related Classes/Methods:
Implements the core algorithm for fitting logistic regression models, suitable for binary or ordinal categorical outcome variables. It calculates coefficients, odds ratios, and statistical significance.
Related Classes/Methods:
Manages the complex workflow for mediation analysis, which involves multiple regression steps to determine direct, indirect, and total effects of an independent variable on a dependent variable through a mediator. It integrates results from basic regression models and applies bootstrapping for robust inference.
Related Classes/Methods:
A utility component responsible for extracting standardized point estimates (e.g., regression coefficients) from the results of various regression models, ensuring consistency in output across different regression types. It directly calls regression models to obtain necessary coefficients.
Related Classes/Methods:
Calculates the relative importance of predictors within a regression model, providing insights into the unique contribution of each independent variable to the model's explanatory power.
Related Classes/Methods:
Computes p-values based on bootstrap confidence intervals, offering a non-parametric approach to statistical inference, particularly useful when assumptions for traditional parametric tests are not met.
Related Classes/Methods:
Calculates bias-corrected confidence intervals, a refinement often used in bootstrapping to improve the accuracy and reliability of interval estimates, especially in complex models like mediation. It processes bootstrap distributions and sample points to derive confidence intervals.
Related Classes/Methods: