Box-Cox Transformation Explorer
Overview
The Box-Cox Transformation Explorer helps you find the optimal transformation to normalize your skewed data. Automatically calculate the best lambda parameter via maximum likelihood, manually explore different transformations with an interactive slider, and visualize before/after distributions with histograms and Q-Q plots. See common transformations like log (λ=0), square root (λ=0.5), and square (λ=2), and compare skewness before and after transformation.
Tips
- The optimal lambda maximizes the log-likelihood, making the transformed data as close to normal as possible
- λ = 1 means no transformation needed (data is already approximately normal)
- λ = 0 corresponds to log transformation, useful for right-skewed data like income or reaction times
- λ = 0.5 (square root) moderately reduces right skew and is more interpretable than arbitrary lambdas
- Box-Cox requires all values to be positive; add a constant if you have zeros or negative values
- After transforming, remember to back-transform predictions for interpretation in the original scale
- Use Q-Q plots to verify normality improvement - points should align better with the diagonal after transformation