[Kaggle Extra Study] 18. Types of Correlation Analysis

There are several types of correlation analysis:

1. Pearson Correlation
   
   • Most commonly used method
   • Measures linear relationships
   • Suitable for continuous variables
   • Values range from -1 to 1
   • Assumes normal distribution
2. Spearman Correlation
   
   • Measures rank-based correlation
   • Captures non-linear relationships well
   • Suitable for ordinal data
   • Less sensitive to outliers
   • Can be used even when data is not normally distributed
3. Kendall's Tau
   
   • Rank-based correlation
   • More suitable for small samples
   • Less sensitive to outliers
   • More complex to calculate than Spearman but can be more accurate

import pandas as pd
import scipy.stats as stats
# Example data
x = [1, 2, 3, 4, 5]
y = [2, 4, 5, 4, 5]
# Pearson correlation
pearson_corr, _ = stats.pearsonr(x, y)
# Spearman correlation
spearman_corr, _ = stats.spearmanr(x, y)
# Kendall's tau
kendall_corr, _ = stats.kendalltau(x, y)

• Choosing the appropriate correlation coefficient is important depending on your data characteristics:
  • Use Pearson for continuous data that follows normal distribution
  • Use Spearman for ordinal data or when data doesn't follow normal distribution
  • Use Kendall's tau for small samples or when rank relationships are important

---

Success is not about achieving a destination, but rather enjoying the journey.
- Max Holloway -