Feature Transformation Techniques

Logarithmic transformation

Use the logarithmic operation to compress the range of values and reduce skewness
Log is essentially finding what the power is needed for the base to obtain the input
Base 2: $log_2(X)=Y$ is finding what should $Y$ be such that $2^Y=X$ , so $log_2(8)=3$ because $2^3=8$
Base 10: $log_{10}(X)=Y$ is finding what should $Y$ be such that $10^Y=X$
Base e or natural log: $ln(X)=Y$ is finding what Y be such that $e^Y=X$

Square Root Transformation

This technique reduces the range of values by taking the square root of each value. It is useful for stabilizing variance.

Box-Cox Transformation