SellerBay

Development and testing of a prototype light industrial demand forecasting system for a distributed manufacturing platform with the ability to determine the cost of purchasing, selling, and required product runs

HOME / OUR WORK / SellerBay

/ about the project

Development and testing of a prototype light industrial demand forecasting system for a distributed manufacturing platform with the ability to determine the cost of purchasing, selling, and required product runs.

/ task

The project is designed to solve the technical problem of developing methods for forecasting demand for light industry goods based on neural network and machine learning technologies and methods for efficient distribution of orders between production facilities based on intelligent recommendation system technologies.

/ Solution

1) Recommendation system provides formation of high-quality recommendations with simultaneous servicing of at least 100 user requests.

2) The accuracy of demand forecasting for a product position is not less than 80%

3) Root-mean-square error of demand for a commodity position is not more than 20%

4) The system provides receiving data of commodity positions and their parameters at a speed of not less than 100 positions per second.

5) Formation of recommendations is carried out in real time mode 24/7.

6) The accuracy of classification of the text description of goods items and recognition of their parameters of the assortment is at least 95%.

7) The speed of formation of recommendations on the selection of goods items is not more than 5 minutes.

/ technologies

Kubernetes, React JS, JupyterNotebook, VS code.
Tensorflow, Keras, numpy, scikit-learn, recurrent neural networks, LSTM. Djangо, Postgress SQL