Analyses

Random forest regression

Bagged trees for a numeric outcome, with holdout R² and RMSE.

Edit on GitHub

When to use it

Use random forest regression when the outcome is numeric and you want a holdout R², RMSE, and feature importance without a linear model. The classification version is random forest classification. A linear fit of the same columns is linear regression.

Assumptions

Outcome and features are numeric. Holdout fraction 0.25, 100 trees, min_samples_leaf 5, no depth cap, random_state=0. HTTP and chat dispatch both honour test_fraction and min_samples_leaf.

Running it in Tensr

ML → Regression → Random Forest Regression. In chat: “Random forest regression of score on hours and anxiety.”
Target is the numeric outcome. Features need at least one numeric column.

Options

Prop

Type

Reading the output

Exam score from hours and anxiety on that same 240-row sample. 100 trees, min_samples_leaf 5, no depth cap. Both slopes were built steep, so holdout R² should be positive.

VariableImportance
hours0.845
anxiety0.155

Holdout R² = 0.721, RMSE = 4.733, n = 240, min_samples_leaf = 5. Hours and anxiety were built with steep slopes, so a positive holdout R² is the expected reading.

Reporting (APA 7)

Random forest regression of exam score on hours and anxiety, n = 240, holdout R² = 0.721, RMSE = 4.733.

Coming from SPSS

SPSS has no random-forest regression dialog. Tensr’s item is on the ML menu, Regression → Random Forest Regression.

Linear regression is the parametric counterpart. Gradient boosting can be switched to regression mode.