R-Squared and Error of Predictions
Use R² and root mean squared error compare predictions with observed values.
Your numbers
R-squared
R² fit negative if predictions poor. This result no adjust for number of parameters wey model fit.
Compare scenarios
You fit keep up to three snapshots for this tab. Loading one go replace the present inputs.
Formula
Use R² = 1−SSE/SST and RMSE = √(SSE/n). SSE na sum of squared prediction errors, SST na sum of squared distances from di observations' mean, and n na count. If all observed values equal, di explained-variation ratio no get defined value.
Use R² and root mean squared error compare predictions with observed values.
Example wey show the calculation
- Values wey you observe
- 2, 4, 6
- Values wey model predict
- 2, 3, 5
- R-squared
- 0.75
- Square root of average squared prediction error
- 0.816496580928
Where e fit work and limits
R² fit negative if predictions poor. This result no adjust for number of parameters wey model fit.
We check inputs for missing values and allowed ranges. Result wey pass the finite numeric range no go show. Rounding affect wetin display, while calculation use the model numeric precision.
Check your units and assumptions before you rely on result. Changing label no convert currency or physical unit.
Version of di formula: a6b164ec3731 · Example don get check: