Survey Analysis

About this Project

     This project was built as part of my externship in the OMSA program. It is still being used in production today with many enhancements and delivers key insights to top leadership each year.  

Objective

        Each year our business conducts a massive survey and simultaneously gathers a handful of metrics which help our leaders make informed decisions throughout the year. We often consider these our north star metrics as they align well to our business principles.  

Since these metrics are only gathered yearly, it can be difficult to use them for business decisions throughout the year. Instead, I will work toward understanding which measurable metrics that we currently monitor at the monthly level correlate strongest toward the north star metrics to serve as proxy metrics. I will do this by creating a feature importance ensemble and comparing the results from several different algorithms. 

Also in this survey are over one million short response questions. This can be difficult for our leaders to consume, even at the lowest level of leadership it is just about 500 responses. I will use GenAI techniques to analyze the sentiment of these comments as well as categorize them into meaningful topics. I will then also use GenAI models to summarize these topics into bullet points so that our leadership can stay informed.