Classical statistics for small analyses: descriptive statistics, the Student t, normal, chi-square and F distributions, t-tests with confidence intervals and effect sizes, Pearson correlation with a p-value, and an ordinary least squares fit with the full summary table (standard errors, t statistics, p values, confidence intervals, F test, R-squared, AIC and BIC). Compile with -lib stats,gen_collect; it needs nothing from System.ML.
No call here ends the program on bad input. A function that returns a Float returns NaN, and a function that returns a result object returns one whose IsValid() is false and whose GetError() says why. Design and verification notes: docs/AI.md, "Statistics".
TTest
t-tests. Every test takes an optional confidence level (default 0.95) for the interval on the mean difference. All return a TTestResult; none can end the program.
Failures (IsValid() false): Nil input, a sample too short (one-sample and paired need 2 values, two-sample needs 2 per group), unequal lengths for the paired test, NaN or infinity in the data, a level not strictly between 0 and 1, and zero variance (a standard error of 0 makes t undefined).
Example
a := [4.1, 5.2, 6.3, 5.8, 4.9, 5.5];
b := [6.0, 7.1, 6.8, 7.4, 6.5, 7.0];
r := TTest->Welch(a, b);
r->ToString()->PrintLine();
if(r->IsValid() & r->GetP() < 0.05) {
"means differ, d = {$r->GetCohenD()}"->PrintLine();
};Operations
OneSample # function
One-sample t-test of the mean of x against a hypothesised mean, at the 95% level.
function : OneSample(x:Float[], mu:Float) ~ TTestResultParameters
| Name | Type | Description |
|---|---|---|
| x | Float | sample, at least 2 values |
| mu | Float | hypothesised mean |
Return
| Type | Description |
|---|---|
| TTestResult | result; the difference is mean(x) - mu |
OneSample # function
One-sample t-test of the mean of x against a hypothesised mean.
function : OneSample(x:Float[], mu:Float, level:Float) ~ TTestResultParameters
| Name | Type | Description |
|---|---|---|
| x | Float | sample, at least 2 values |
| mu | Float | hypothesised mean |
| level | Float | confidence level of the interval, strictly between 0 and 1 |
Return
| Type | Description |
|---|---|
| TTestResult | result; the difference is mean(x) - mu |
Paired # function
Paired t-test, at the 95% level.
function : Paired(x:Float[], y:Float[]) ~ TTestResultParameters
| Name | Type | Description |
|---|---|---|
| x | Float | first measurements |
| y | Float | second measurements, same length as x, at least 2 |
Return
| Type | Description |
|---|---|
| TTestResult | result; the difference is mean(x - y) |
Paired # function
Paired t-test: a one-sample test of the differences x - y against 0. Cohen's d is d_z, the mean difference over the standard deviation of the differences.
function : Paired(x:Float[], y:Float[], level:Float) ~ TTestResultParameters
| Name | Type | Description |
|---|---|---|
| x | Float | first measurements |
| y | Float | second measurements, same length as x, at least 2 |
| level | Float | confidence level of the interval, strictly between 0 and 1 |
Return
| Type | Description |
|---|---|
| TTestResult | result; the difference is mean(x - y) |
TwoSample # function
Student's two-sample t-test (equal variances assumed), at the 95% level.
function : TwoSample(x:Float[], y:Float[]) ~ TTestResultParameters
| Name | Type | Description |
|---|---|---|
| x | Float | first sample, at least 2 values |
| y | Float | second sample, at least 2 values |
Return
| Type | Description |
|---|---|
| TTestResult | result; the difference is mean(x) - mean(y) |
TwoSample # function
Student's two-sample t-test (equal variances assumed, pooled variance, df = n1 + n2 - 2).
function : TwoSample(x:Float[], y:Float[], level:Float) ~ TTestResultParameters
| Name | Type | Description |
|---|---|---|
| x | Float | first sample, at least 2 values |
| y | Float | second sample, at least 2 values |
| level | Float | confidence level of the interval, strictly between 0 and 1 |
Return
| Type | Description |
|---|---|
| TTestResult | result; the difference is mean(x) - mean(y) |
Welch # function
Welch's t-test (unequal variances, Welch-Satterthwaite degrees of freedom), at the 95% level.
function : Welch(x:Float[], y:Float[]) ~ TTestResultParameters
| Name | Type | Description |
|---|---|---|
| x | Float | first sample, at least 2 values |
| y | Float | second sample, at least 2 values |
Return
| Type | Description |
|---|---|
| TTestResult | result; the difference is mean(x) - mean(y) |
Welch # function
Welch's t-test (unequal variances). Cohen's d and Hedges' g use the pooled standard deviation, as for the Student test.
function : Welch(x:Float[], y:Float[], level:Float) ~ TTestResultParameters
| Name | Type | Description |
|---|---|---|
| x | Float | first sample, at least 2 values |
| y | Float | second sample, at least 2 values |
| level | Float | confidence level of the interval, strictly between 0 and 1 |
Return
| Type | Description |
|---|---|
| TTestResult | result; the difference is mean(x) - mean(y) |