This system has been developed for a major car electrical equipment manufacturer in the Middle East. We were asked to create a precise demand forecasting system so the client can precisely forecasts part demands for its handful of clients in high volume without any spontaneous spikes in demands and unwanted surprises.

We had to think out of the box and not only use the latest advancements in time series trends forecasts using RNN and multi step LSTM we had to employ probabilities, and econometrics to tune the system even further and make it work the way that suits our client.

We also had to digest data not also from their currently available CRM but also we had to consider annual occasions, psychological marketing attempts of factory users in order to take into consideration even the most unexpected parameters that can affect the demand as well.

We were able to create the system in fastest time possible leveraging as much data as possible including IoT devices on their warehouse and transportation infrastructure to monitor and forecast the warehouse products with the accuracy of 80% and we managed to create a system to be an online learning neural network which tunes and adopts throughout its life span.