# regressionStats: calculate prediction performance statistics In environmentalinformatics-marburg/Rsenal: Magic R Functions for Things Various

## Description

this function calculates prediction performance statistics between vectors of predicted and observed values, namely coefficient of determination (Rsq), root mean squared error (RMSE), mean error (ME), mean absolute error (MAE). Users may also create a dotplot visualising the results.

## Usage

 ```1 2``` ```regressionStats(prd, obs, adj.rsq = TRUE, plot = FALSE, method = "pearson") ```

## Arguments

 `prd` numeric vector of predicted values `obs` numeric vector of observed values `adj.rsq` logical, whether to return adjusted r-squared. Defaults to TRUE `plot` logical, whether to produce a visualisation of the results. Defaults to FALSE `method` character. Method to use for correlation. See `?cor.test` for details.

## Value

If `plot = FALSE` (the default), a data frame. If `plot = TRUE`, a list with components `stats` - data frame and `plot` - a trellis plot object.

## Author(s)

Tim Appelhans, Hanna Meyer

## Examples

 ``` 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15``` ```## create predictions with high accuracy (identical mean), ## but low precision (sd double that of observations). Hence, ## ME should be close to zero and RMSE close to ten. pred_vals <- sort(rnorm(1000, 200, 20)) # sorting ensures high Rsq obs_vals <- sort(rnorm(1000, 200, 10)) ## with plot = TRUE result <- regressionStats(pred_vals, obs_vals, plot = TRUE) result\$stats result\$plot ## with plot = FALSE result <- regressionStats(pred_vals, obs_vals, plot = FALSE, adj.rsq = FALSE) result ```

environmentalinformatics-marburg/Rsenal documentation built on Sept. 23, 2018, 5:17 a.m.