Table of Contents

Enum TestStatisticType

Namespace
AiDotNet.Enums
Assembly
AiDotNet.dll

Represents different types of statistical tests used to evaluate hypotheses and determine significance in data analysis.

public enum TestStatisticType

Fields

ChiSquare = 0

A statistical test used to determine if there is a significant association between categorical variables.

For Beginners: The Chi-Square test helps us understand if two categorical characteristics are related.

Imagine you want to know if ice cream flavor preference (chocolate, vanilla, strawberry) is related to gender. The Chi-Square test compares the actual distribution of preferences across genders with what we would expect if there was no relationship.

When to use it:

  • When your data falls into categories (like yes/no, red/blue/green, etc.)
  • When you want to know if one categorical variable is related to another
  • When you have counted data (frequencies) rather than measurements

Example: Testing if treatment type (medication A, B, or placebo) is related to recovery outcome (recovered/not recovered).

FTest = 1

A statistical test that compares the variances of two or more groups to determine if they are significantly different.

For Beginners: The F-Test helps us compare the spread or variability between different groups.

Imagine comparing the test scores from three different teaching methods. The F-Test can tell us if one method produces more consistent results (less variability) than others.

The F-Test is also the foundation of ANOVA (Analysis of Variance), which compares means across multiple groups.

When to use it:

  • When comparing more than two groups
  • When your data is numerical (like heights, weights, scores)
  • When you want to know if groups differ in their average values

Example: Testing if three different fertilizers produce different average crop yields.

MannWhitneyU = 3

A non-parametric test that compares two independent samples without assuming they follow a normal distribution.

For Beginners: The Mann-Whitney U test is like a T-Test but works when your data doesn't follow a nice, neat pattern.

Imagine comparing customer satisfaction ratings (1-5 stars) between two restaurants. Since ratings are often skewed (not following a bell curve), the Mann-Whitney U test is more appropriate than a T-Test.

When to use it:

  • When comparing two independent groups
  • When your data doesn't follow a normal distribution
  • When your data is ordinal (has a natural order but not equal intervals)
  • When you have outliers that might skew results

Example: Comparing pain relief scores (on a scale of 1-10) between two different treatments.

PermutationTest = 4

A resampling-based test that repeatedly shuffles observed data to determine if patterns are statistically significant.

For Beginners: The Permutation Test is like shuffling a deck of cards many times to see how likely a particular arrangement is.

Imagine you have test scores from students who studied using two different methods. You mix all scores together and randomly reassign them to the two methods thousands of times. If the original difference between methods is larger than what you typically see in these random reassignments, it suggests the difference is significant.

When to use it:

  • When traditional tests' assumptions aren't met
  • When you have small sample sizes
  • When you want to avoid making assumptions about your data's distribution
  • When you need a flexible approach for complex data

Example: Testing if gene expression patterns differ between healthy and diseased tissue samples.

Note: Permutation tests are computationally intensive but very flexible and powerful.

TTest = 2

A statistical test used to determine if there is a significant difference between the means of two groups.

For Beginners: The T-Test helps us decide if two groups have different averages.

Imagine comparing the heights of men and women. The T-Test tells us if the difference in average height between the two groups is statistically significant or could have happened by chance.

When to use it:

  • When comparing exactly two groups
  • When your data is numerical (like heights, weights, scores)
  • When your data approximately follows a normal distribution (bell curve)

Example: Testing if a new medication affects blood pressure by comparing before and after measurements.

Remarks

For Beginners: Statistical tests help us decide if patterns we see in data are real or just due to chance.

Think of statistical tests like different tools in a toolbox - each one is designed for specific situations:

Imagine you're trying to determine if a coin is fair (50% chance of heads). You could flip it 100 times and count how many heads you get. If you get exactly 50 heads, it seems fair. But what if you get 55 heads? Or 60? At what point do you decide the coin is unfair? Statistical tests give us mathematical ways to make these decisions based on probability rather than just guessing.

Different tests are designed for different types of data and questions, just like you'd use different tools for different home repair jobs.

These tests calculate a "p-value" - the probability that the pattern you observed could happen by random chance. A small p-value (typically < 0.05) suggests the pattern is statistically significant and not just random.

Edit this page