PairedTTest<T>
Models & Types · AiDotNet.Evaluation.Statistics
Paired t-test: compares means of two related samples.
Implements: IPairedTest<T>
For Beginners
Use this test when you have paired observations, such as:
- Before/after measurements on the same subjects
- Same dataset evaluated by two different models (paired by sample)
- Cross-validation fold results for two models on the same folds
The test determines if there’s a significant difference between the pairs.
How It Works
Assumptions:
- The differences between pairs are approximately normally distributed
- Pairs are independent of each other
- Data is continuous (or at least ordinal with many levels)
Constructors
| Constructor | Summary |
|---|---|
PairedTTest() |
Properties
| Property | Summary |
|---|---|
string Description { get; } |
|
string Name { get; } |
Methods
| Method | Summary |
|---|---|
StatisticalTestResult<T> Test(T[] sample1, T[] sample2, double alpha) |