Data Scientist
Current Created a look alike model using logistic regression with interaction effects for a retail company to acquire new households who are predisposed to buying their products. The model captured 3.85 times more leads than a random targeting campaign Constructed a program in Python that collapsed the levels of categorical variables based on the reduction in Chi-square test between the nominal variable and the target thereby making it easy to include nominal variables with many levels in model Developed a process in R which compares targeted segments against a few hundred variables and produces a multi-tabbed Excel spreadsheet complete with comparative distributions Calculated market penetration at zip2 level & 80% footprint to identify market opportunities and identified target households who are likely utilize their services Developed, documented and published locally several tools and techniques which are now common practices Created client ready presentations to include: objectives, methodology, results, executive summary and recommendations