⚙️ A microservice for water usage forecasts with support for on-demand and preconfigured forecasting algorithms
Note
This microservice replaces the two-part microservice to minimize the used technologies in the WISdoM platform. All features present in the two-part microservice will be ported.
The only feature that will not be ported is the communication using AMQP since this technology will be discontinued in the WISdoM platform.
This microservice allows users to use one of the following pre-built algorithms to generate water usage forecasts:
- Linear Regression
- Polynomial Regression (up to the 5th degree)
- Logarithmic Regression
Note
The storage and usage of custom forecasts is currently under active development and may not be usable or show unexpected behaviour.
Note
The addition of more languages is planned but for the time being only python is supported by default.
| Language | Version |
|---|---|
| Python | v3.10 |
Important
The on-demand forecasts are currently a WIP since there are still some issues with script isolation. Therefore, on-demand forecasts will not be available at this time.
The service accepts custom on-demand written forecasting models and allows a fast testing and validation of new forecasting models. These on-demand forecasts are executed in a new docker container.
To allow the usage of preconfigured forecasts in addition to the already pre-built algorithms, users may use the upload endpoint in this microservice as it is documented.