We addressed the client’s challenge in a 5-step process:
- We used an electronic design automation of representative stores to hypothesize the structural variables in the analysis.
- We built classiﬁcation tree to identify key structural variable impacting the sales in the stores.
- The outliers stores were then removed through clustering of the existing stores based on key structural variables such as population density, big box retailers, and eateries, controlling the impact of executional variables.
- This was followed by a regression to understand key drivers for each cluster.
- Finally, we prioritized the variables based on their inﬂuence on store performance.
- The web application designed for the client provided an intuitive visualization layer
- It provided the client the ﬂexibility to apply any prospect location, in return giving him a gross addition forecast in near real-time.
- The solution fetches real-time information from diverse platform such as Google and Yelp