Start Forecast & Training

Forecast Details

Element	Title	Details
[A]	General Configuration	Parent block with all settings for the forecast model.
[A1]	Asset	The selected asset whose data is used for the forecast.
[A2]	Target Attribute	Attribute for which the forecast is calculated.
[A3]	Feature Attributes	List of additional attributes (minute, hour, weekday...); they serve as input features for the model.
[A4]	Forecast length	Number of time steps to be predicted into the future.
[A5]	Context length	Length of the historical context window that the model considers for each prediction.
[A6]	start/stop training	Switch to start/stop the training run.
[A7]	start/stop forecasting	Switch to activate/deactivate the ongoing forecast.
[B]	Info	Status panel with runtime information.
[B1]	Forecast Status	Shows whether the forecast is active/inactive and, if active, in what state the forecast is.
[B2]	Training Status	Shows the status of the training active/inactive and in what state the training is.
[B3]	DOC	Link to the online documentation of the Forecast app.
[C]	Forecast Chart	Chart with actual and forecast values.
[C1]	Legend – Actual value	Identifies the measured sensor data (Actual humidity).
[C2]	Legend – Forecast	Identifies the calculated predictions (Forecast humidity).
[C3]	Forecast Line	Shows the prediction as a line in the chart.

Start Forecast & Training

The Forecast App allows you to activate model training and the ongoing forecast independently of each other. Both processes run in parallel in the background but follow different tasks and procedures. The following sections explain the functionality, dependencies, and recommended procedure.

Start Training

Goal: Trains a new LSTM model based on the currently available historical data.

Procedure:

• The training process is started via the Train [A6] switch.

• The model is trained with the current settings for forecast_length, context_length, and feature_attributes.

• During training, the best model at any given time (measured by the Validation Loss) is automatically cached.

Important:

• The training status is displayed in the Info panel.

• After successful completion, the message appears: → Training completed successfully. Waiting for enough data to retrain.

Automatic Retrain: After the initial training, the model is automatically retrained as soon as at least 10% more data points are available than in the last run. → Example: If a model was trained on 10 months of data, the next training starts as soon as another month of data has been added.

---

Start Forecast

Goal: Creates ongoing forecasts for new, incoming data points based on the most recently trained model.

Procedure:

• The forecast is activated via the Forecast [A7] switch.

• The app checks whether a trained model is available. Only then can the forecast begin.

• New forecast values are automatically generated and saved as an attribute in the asset (see → Output Attribute of the Forecast).

Note: If no model is available yet (e.g., after the initial creation), no forecast can be made. In this case, the status will show: → Waiting for model to be trained.

---

Recommendation for the Order

To achieve reliable results, it is recommended to:

1. First start the training

2. Then activate the forecast after successful completion

This way you avoid the first predictions being based on an untrained or poorly converged model.

---

Technical Terms in the Status Area

Term	Meaning
Epochs	The number of complete training runs over the entire dataset.
Validation Loss (val_loss)	Error metric on the validation data. Shows how well the model generalizes. The lower, the better.

These values help to assess the quality of the training process.

---

Output Attribute of the Forecast

When starting the forecast or the training process, the Forecast App automatically creates a new attribute in the selected asset. This attribute contains the calculated forecast values and is available system-wide.

Naming scheme: [target_attribute]_forecast_[forecast_length]

Example: If the temperature attribute is predicted with a forecast_length of 12, the name of the automatically generated output attribute is: → temperature_forecast_12

Note: This attribute can be used anywhere in Eliona—e.g., in dashboards, rules, visualizations, or the Calculator. It behaves like a regular asset attribute and is automatically updated with the latest forecast values.