BootstrapTest<T>
Models & Types · AiDotNet.Evaluation.Statistics
Bootstrap-based hypothesis test for comparing two models.
Implements: IStatisticalTest<T>
For Beginners
Bootstrap testing uses resampling to test significance:
- Repeatedly samples with replacement from your data
- Computes the statistic of interest on each sample
- Builds a distribution to test your hypothesis
- No assumptions about the underlying distribution
How It Works
Advantages:
- Works for any metric, not just specific test statistics
- No distributional assumptions
- Can compute confidence intervals for any statistic
Constructors
| Constructor | Summary |
|---|---|
BootstrapTest(int numBootstraps, int? randomSeed) |
Initializes the bootstrap test. |
Properties
| Property | Summary |
|---|---|
bool IsNonParametric { get; } |
|
bool IsPaired { get; } |
|
string Description { get; } |
|
string Name { get; } |
Methods
| Method | Summary |
|---|---|
StatisticalTestResult<T> Test(T[] scores1, T[] scores2) |
Tests if two sets of scores differ significantly using bootstrap. |
ValueTuple<double, double> ComputeCI(T[] scores1, T[] scores2, double confidenceLevel) |
Computes bootstrap confidence interval for the mean difference. |