Build a forast model / artificial intelligence
Budget: $30 – $250 USD
I'll need a model to forecast discharge measurements in a river paset on psat dichscharge measurements and rainfall predictions. Both series have 3h resolution and contain approx. 1500 values.
For the task, the darts-library in Python shall be used. See https://github.com/unit8co/darts and, e.g., https://medium.com/unit8-machine-learning-publication/time-series-forecasting-using-past-and-future-external-data-with-darts-1f0539585993. Use one of the models that is multivariate (e.g., BlockRNNModel, NBEATSModel).
The tasks consists in two parts:
- Model creation, calibration and validation
- Provide a python function that can be run every three hours to get the next prediction without requiring model re-calibration every three hours.
For the task, the darts-library in Python shall be used. See https://github.com/unit8co/darts and, e.g., https://medium.com/unit8-machine-learning-publication/time-series-forecasting-using-past-and-future-external-data-with-darts-1f0539585993. Use one of the models that is multivariate (e.g., BlockRNNModel, NBEATSModel).
The tasks consists in two parts:
- Model creation, calibration and validation
- Provide a python function that can be run every three hours to get the next prediction without requiring model re-calibration every three hours.