2 days work and we did it! Our customer was satisfied. We proved to him that his idea is real and now he has a ready-made solution for the future service.
We received a task to automate refilling for home appliances. Like a vending machine or water bottles, fuel or any resource. The problem was to predict the time of next refilling (or when the resource will be almost empty).
How did we solve this question?
We transformed the task of determining the balance in the task of determining how many hours left to zero.
We used a recurrent neural network, RNN, tried Simple RNN, GRU, and LSTM
This graph shows how accurate this engine is.
The blue line is the actual end of resources in hours, orange line – Engine forecast.
As you can see, our engine has shown good results.
Also, it is possible to automatically order resources when they are needed. And the user only needs to confirm.
Hence, if this question interests you, please, send us an e-mail firstname.lastname@example.org and receive consultation according to this project.