Behind the scene
The idea originated when we were thinking of what could be a burning need for us if we were to run a small to medium sized business. As a shop owner with an elegant looking Clover Station loaded with cool applications, I would love to have an app which could tell me approximately how many items I would be selling and what the potential traffic in my store could be.
We started with the historical transaction data that was available for a merchant in the Clover database. As we were crunching the numbers we soon realized that both the leaderboard of predicted sales as well as "Busyness Index" (We coined this term as it pretty much indicates how busy a store is going to be) were based on quite a lot of variables. We started to write down an exhaustive list of variables such as "Weather", "Impact on Promotions", "Impact on Holidays", "Impact of events happening in neighborhood" etc. For want of time, we decided to just pursue with including the weather variable in our fully working MVP solution.
What we have done in a nutshell is marry the historical weather data(simulated) available for the locality of a particular business with the historical transaction data. The question we asked ourselves as we were developing the algorithm was "If tomorrow is going to be Rainy(Compared to a sunny today), will the coffee sales in my store go up(Compared to an average day) because a lot more people might want something hot to drink on a cold morning?" I think we got an extremely convincing answer when we crunched the numbers.
The feature that we are really proud of(We love all the features!!) is definitely the leaderboard of items that I would potentially be selling during my peak hours. We feel that with more data and more variables the precision would only get better and thereby helping a store owner reduce wastage and also manage store staffing in a smart way.