, , 1900
We assessed the competitive strength of 3 major ATM vendors using spatial ATM network and the demographics of California. We visualized feature importance using Tableau and clustered the ATM locations by utilizing k-means approach, and reverse engineered the feature weights by applying Random Forest Classifier on the clustered data for each county. I built a county specific linear regressor to model the revenue of each ATM, and enhanced the whole California state linear model by adding county-specific linear model to estimate the annual revenue generation of each ATM location.