January 15, 2021
For large datasets with significant variance, it becomes difficult for one model to capture all the underlying patterns. In such cases it is advisable to break the dataset into homogenous subsets and then build separate model for each cluster. When we say homogenous, we mean homogeneity in terms of independent variables.
Thus we use clustering algorithms to split dataset into multiple homogenous subsets and build models customized to that subset.
by : Monis Khan
Quick Summary:
For large datasets with significant variance, it becomes difficult for one model to capture all the underlying patterns. In such cases it is advisable to break the dataset into homogenous subsets and then build separate model for each cluster. When we say homogenous, we mean homogeneity in terms of independent variables. Thus we use clustering […]