Рет қаралды 8,361
Apart from missing or outlier treatment, Dimensionality reduction, one-hot encoding, Data Transformation is an important part of Data pre-processing stage. If done effectively, this leads to improved model performance.
There are many such techniques like - Log or power transformation, Winsorization or clipping, Unit Vector scaling, etc. Each of them have mathematical basis which makes it more popular in one area than other.
This video talks about two popular techniques of Data Transformation - Normalization & Standardization. Both of them can easily be implemented using popular tools like Python, R, etc.
For similar topics, visit - www.datarlabs.com/