- There are large amount of food waste due to distribution problem. To tackle this problem, we train a model that will predict how much ingredient to buy for the shop to avoid food wastage.
What it does
- This application able to predict how much ingredients to buy on that particular date. By having a good prediction of the amount of ingredients buy, we are able to reduce waste by reducing the amount of ex cess ingredients.
How we built it
- We us an online model called Prophet to train the data by having the previous date data. It is a forecasting time series model and it help us to predict the quantity purchased at December 2020.
Challenges we ran into
-In this project, we realise that the sales data and the items purchase do not have any correlation, hence we are unable to use some machine learning techniques such as linear regression and decision tree regression.
- The differences between the predicted and actual purchase is rather large is due to the sudden drop of purchase and sales due to Covid-19. During that period especially the beginning of 2020, the amount of ingredients' purchases have dropped dramatically and cause the model to distort a little when training the data.
- We will require more variables such as the weather, the inventory, the perishability of ingredients, and the chef's personality to generate more accurate predictions
Accomplishments that we're proud of
- We are able to train another model that will predict the future sales and give a suggestion to the outlet owner to buy lesser amount of food to avoid food wastage
What we learned
- During this hackathon, we have learned how to apply time series forecast model to the datas. Besides, we learned that we need to find correlation between the input and output before training as we realised that when we apply linear regression to the data, the accuracy is very low.
What's next for Eh Eye
- We are planning to add some more data such as the Covid-19 cases in Singapore and the oil prices to find whether there is any correlation to the amount of ingredients' purchases.