Regression model tools

Python scripts for regression models, using the Scikit-Learn framework:

• Diagnostic plots
• Bootstrapped confidence intervals for predictions
• Approximate Shapley values

Diagnostic plots

While ML models do not generally have the same residual distribution assumptions as for classical linear regression, there is still value in examining residual plots.

import lightgbm as lgb
import pandas as pd
from sklearn.datasets import load_boston
from sklearn.model_selection import train_test_split
from regression_diagnostics import RegressionDiagnostics
import warnings
warnings.filterwarnings('ignore')

# Load the boston house-prices dataset and fit a regression model
boston = load_boston()

X = pd.DataFrame(boston["data"], columns=boston.feature_names)
y = boston.target
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

# Fit model
lgb_model = lgb.LGBMRegressor()
lgb_model.fit(X_train, y_train)

# Generate diagnostic plots
diagnostics = RegressionDiagnostics(lgb_model)
diagnostics.fit(X_test, y_test)

# Fitted values against actual values
diagnostics.fitted_actual()

# Residuals against fitted values
diagnostics.residuals_fitted()

# Histogram of residuals
diagnostics.hist_residuals()

# QQ plot of residuals
diagnostics.qq_plot()

Bootstrapped confidence intervals for predictions

A script to generate local bootstrapped confidence intervals for predictions using observed residuals for k nearest neighbours in a reference data set. Increasing the value of k obtains results closer to a global error interval.