Inspiration
It is a great headache for e-commerce businesses that how much stock should be present for optimal revenue.
What it does
Our aim is to do demand forecasting so that optimized stock of each product will be present.
How we built it
First of all, we built the dataset from the provided data by the code fest organizer. Actually, the data provided was very irrational. So analyzed, cleaned and manipulate the provided data to get our desired data.
Then we used AI and specifically BLSTM to train the model and get prediction.
Challenges we ran into
The most challenging part was to analyze the dataset and cleaned and manipulate to get the good dataset.
Accomplishments that we're proud of
We consider that we have done networking with other fellows and professionals. On technical side, we worked for first time with such a dataset and cleaning. Finally, we built the model and got accuracy of 90% plus.
What we learned
We learned about: what are the real-world actual problems.
What's next for Demand Forecaster
We can work on accuracy improvement and dataset enhancement which is more realistic.
Built With
- colab
- numpy
- python
- scikit-learn
- tensorflow
Log in or sign up for Devpost to join the conversation.