performance: Evaluate Regression Model Performance using Various Metrics

In DMTL: Tools for Applying Distribution Mapping Based Transfer Learning

Description

This function produces the predictive performance for a regression model using various common performance metrics such as MSE, R-squared, or Correlation coefficients.

Usage

performance(y_obs, y_pred, measures = c("NRMSE", "NMAE", "PCC"))

Arguments

y_obs	Observed response values
y_pred	Predicted response values
measures	Performance measures. One can specify a single measure or a vector containing multiple measures in terms of common error or similarity metrics. The available options are roughly divided into 3 categories - "MSE", "RMSE", "NRMSE" for mean squared error, root mean squared error, and normalized root mean squared error, respectively. "MAE", "NMAE" for mean absolute error, and normalized mean absolute error, respectively. "PCC", "SCC", "RSQ" for Pearson's correlation, Spearman's correlation, and R-squared, respectively. Defaults to c("NRMSE", "NMAE", "PCC").

Value

A vector containing the performance metric values.

Examples

set.seed(654321)
x <- rnorm(1000, 0.2, 0.5)
y <- x^2 + rnorm(1000, 0, 0.1)
y_fit <- predict(lm(y ~ x))
print(performance(y, y_fit, measures = c("MSE", "RSQ")))

DMTL documentation built on Feb. 18, 2021, 5:06 p.m.