Select the method or formula of your choice.
The Confusion matrix is not present when the splitting method is class probability.
Count
When there are no weights, the counts and the sample sizes are the same.
Weighted count
In the weighted case, the weighted count is the sum of the weights for a
category. Weighted counts round to the nearest whole number. Use the unrounded
weights to calculate percentages and rates. Consider the following simple
example:
| Response level | Predicted level | Weight |
|---|---|---|
| Yes | Yes | 0.1 |
| Yes | Yes | 0.2 |
| Yes | No | 0.3 |
| Yes | No | 0.4 |
| No | No | 0.5 |
| No | No | 0.6 |
| No | Yes | 0.7 |
| No | Yes | 0.8 |
This table provides the following statistics:
| Actual class | Weighted count | Predicted class = Yes | Predicted Class = No | Percent correct |
|---|---|---|---|---|
| Yes | 0.1 + 0.2 + 0.3 + 0.4 = 1 | 0.1 + 0.2 = 0.3 ≈ 0 | 0.3 + 0.4 = 0.7 ≈ 1 | 0.3 / (0.3 + 0.7) ×100 = 30% |
| No | 0.5 + 0.6 + 0.7 + 0.8 = 2.6 ≈ 3 | 0.7 + 0.8 = 1.5 ≈ 2 | 0.5 + 0.6 = 1.1 ≈ 1 | 1.1 / (1.5 + 1.1) × 100 = 42.31% |
| All | 1 + 2.6 = 3.6 ≈ 4 | 0.3 + 1.5 = 1.8 ≈ 2 | 0.7 + 1.1 = 1.8 ≈ 2 | (0.3 + 1.1) / 3.6 × 100 = 38.89% |
True positive rate (sensitivity or power)
False positive rate (type I error)
False negative rate (type II error)
True negative rate (specificity)
