What it does
Create an algorithm that could provide an automated prediction for the purchasing needs.
How we built it
We use the time series forecasting algorithm fbprophet to build the model, then constantly adjust the parameters, optimize the model, and predict the number of future purchases.
Challenges we ran into
- At the beginning we built an LSTM network to predict the number of purchases, but the prediction results were very unsatisfactory.
- Struggling to perform basic preprocessing of the data set. ## Accomplishments that we're proud of Use the model finally completed the prediction. ## What we learned Time series forecasting algorithm, parameter adjustment method，etc. ## What's next for Ebuy Track Perform more complete processing of the data set, and continuously optimize and adjust the network to obtain more accurate prediction results.