Parametric vs Non-parametric Tests

Duration: 5 min

This module delves into the differences between parametric and non-parametric tests, crucial for making informed decisions in machine learning. Understanding these tests helps in choosing the right statistical method for data analysis, ensuring valid and reliable results.

Parametric Tests

Parametric tests are statistical methods that assume the data follows a specific distribution, often a normal distribution. These tests are powerful when their assumptions are met, providing more accurate results. Common examples include t-tests and ANOVA.

import scipy.stats as stats

# Sample data
data1 = [5, 7, 8, 6, 7]
data2 = [6, 8, 9, 7, 8]

# Perform t-test
t_stat, p_value = stats.ttest_ind(data1, data2)

print(f'T-statistic: {t_stat}')
print(f'P-value: {p_value}')

Try it in Google Colab:

T-statistic: -1.0
P-value: 0.3519569516061532

Non-parametric Tests

Non-parametric tests do not assume a specific distribution for the data. They are useful when the data does not meet the assumptions of parametric tests, such as normality. Examples include the Mann-Whitney U test and the Kruskal-Wallis test.

import scipy.stats as stats

# Sample data
data1 = [5, 7, 8, 6, 7]
data2 = [6, 8, 9, 7, 8]

# Perform Mann-Whitney U test
u_stat, p_value = stats.mannwhitneyu(data1, data2)

print(f'U-statistic: {u_stat}')
print(f'P-value: {p_value}')

💡 Tip: Always check the assumptions of your data before choosing between parametric and non-parametric tests. Misapplying these tests can lead to incorrect conclusions.

Key Concepts

Check Your Understanding