Table 2 Performance of the prediction models for percentage mammographic density (PMD)
From: Predicting mammographic density with linear ultrasound transducers
Prediction model | Apparent measures | Cross-validated measures | ||
|---|---|---|---|---|
MSE | R2 | MSE | R2 | |
M0: Null model | 0.0396 | 0.000 | 0.0400 | − 0.021 |
M1: Linear regression | 0.0281 | 0.289 | 0.0391 | − 0.002 |
M2: Linear regression with variable selection | 0.0285 | 0.280 | 0.0386 | 0.011 |
M3: Cubic splines | 0.0187 | 0.528 | 0.0370 | 0.053 |
M4: Cubic splines with variable selection | 0.0200 | 0.494 | 0.0346 | 0.114 |
M5: Lasso | 0.0294 | 0.256 | 0.0367 | 0.061 |
M6: Ridge regression | 0.0308 | 0.222 | 0.0368 | 0.060 |
M7: Random forest | 0.0049 | 0.876 | 0.0291 | 0.255 |