The Measurement Minute by Gary Angel

The hardest part of most analytics projects is the data wrangling – and so far that’s proven to be the case when it comes to building a predictive queue model to incorporate into our DM1 platform. I’m working with data from multiple clients and while the core data around things like store occupancy, time in store and entrance was quite easy to put together, building the line-depth and register open files was more challenging. That data can’t be lifted directly from the interface because in the platform we take advance of historical data points (like who ended up reaching the queue and who was in a group) that won’t be available or available in the same way for a real-time model. Today’s Minute explains what was missing and how I went about filling in the data gaps to create the MVP version of a predictive queue data file for model building.

Podcast: Play in new window | Download

Subscribe: RSS