Predict water consumption per 30 min according to historical data

Hey, It sounds like a Time-series forecasting task. I have some experience about this, I think ARIMA model will be good for this forecasting. I can help you implement ARIMA model or something else basic your need, and I will help you integrate it to your .NET code.

I am a data scientist with PhD. I've studied complex phenomena using theoretical and data analysis, including time series analysis. Your project concerns about the very time series analysis. I have much experience on data analysis of animal behavior. An example is pseudo-periodic movement of snake like animals. Your data would have seasonal periodic dynamics too. Thus I think mathematical modeling with some periodicity be appropriate rather than neural networks. But, I can test some recurrent type neural nets, like GRU, as well. To be noted, I use python and so the python code will be the final product. I also show the performance of the algorithm based on the past data.

I can build a regression neural network using Keras framework with python for this. What is the significance of the Date column - Is there a significance between weekday / Weekend day ? Or else, we can just use the date to figure out the hours taken before the next consumption. As far as integration with .Net concerned, I have no clear idea, may be is it okay if you can run a python server (flask ??) where you can send a request and get a response ?

Hi, I am an expert .NET developer interested in your project for predicting water consumption. I work in the utility industry, for over a decade, and have direct experience that will be of great use to you. I have developed solutions for estimation, historical usage and many other consumption processes for measurement and billing of utilities including water, gas and electricity. I would be glad to discuss this in more detail and collaborate in this project with you. Thanks, Jose Cruz