One of the most tedious tasks for a bar manager is inventory management. If there was a way to easily know when inventory was running low and quickly reorder it would streamline the bar management process.
What it does
TapSwitch is an automated reordering suggestion system based on expected consumption from historical performance as well as external factors.
Using historical information, we assign a priority score based on amount of beer poured per day, weather external factors (some beers sell better during some weather) in upcoming forecast, and organizational priorities (e.g. Sales). After the scores are aggregated, TapSwitch will pull the top 5 beers and suggest them to the user to reorder and allow for text/app/call reordering.
How we built it
We used the Quality/Location and Draught tables to find consumption by bar by brand. We then analyzed the consumption against weather to find the preferred beers based on weather. Then parse through the priority score that is assigned at the two steps above and if the value is >2 it is presented to the user as a suggested beer. We only take the top 5 beers at the time so manual manipulation will help with capturing beers that fall lower in the priority.
Challenges we ran into
The dataset formatting gave us a fair amount of issues given that there were issues either joining tables due to lack of keys or even reading them (one file we found was ; delimited). The formatting ended up taking a bulk of our time as well as many of us learning new tools for the first time (e.g. Pandas in Python)
Accomplishments that we're proud of
We're incredibly proud of how we chose to assign priority scores as this came from hours of poking around the dataset. We think that this is a novel concept to promote push marketing as well as get a better grip on production.
What we learned
Always look at the dataset indepth before processing.
What's next for TapSwitch
We'd like to:
- Build in prediction triggers for major events by region
- Order Quantity Prediction
- Prediction by venue type