⚙️ 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.
| Language | Version |
|---|---|
| Python | v3.10 |
| R | v4.3.1 |
Important
To support on-demand forecasts, the microservice needs access to the Docker Host. Read more here
To allow a dynamic creation and adaptation of forecasting algorithms, the service allows forecast algorithms written in R or Python. When receiving an on-demand custom forecast request, the script sent in the request body will be parsed and put into a container. This container will then be built with the needed requirements and will then be executed as a one-off service.
Read more here
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.