At a certain company O in Chiyoda-ku, a number of hungry staffs are coming up to the cafeteria for a boxed lunch at lunchtime to supply energy, which is necessary to work actively from the afternoon. "Is it cost-effective?", "I have no time going out for lunch.", "My favorites are in the box. I will take it.". The boxes are going off one by one as staff thoughts crossing each other. Aside to that, when they remain unsold, it is not friendly to either the environment or the seller's finances. Even when they go out of stock, it might be an opportunity loss for the seller and time loss for the staff coming to the cafeteria. In this way, setting up a production plan based on the accurate demand forecast is considered to be important for the increase in sales of boxed lunch and waste minimization. Therefore, Opt presents you a challenge to create a model that predicts the appropriate number of boxed lunch to produce based on the various variables such as day of week and menu.
Available data period is from November 18, 2013 to November 30, 2014 except for Saturdays, Sundays, and holidays. Train set: November 18, 2013 to September 30, 2014 Test set: October 1, 3014 to November 30, 2014
Scored on the RMSE (Root Mean Squared Error)
Details : https://signate.jp/competitions/24
- Clone fastai via "git clone`https://github.com/fastai/fastai"
- Install missing dependent packages as per https://medium.com/@GuruAtWork/fast-ai-lesson-1-7fc38e978d37