YousefMohassab opened a new issue, #16958:
URL: https://github.com/apache/iotdb/issues/16958
## Feature Request
Add **Prophet** (Facebook/Meta) and **StatsForecaster** to the built-in
forecast models in IoTDB AINode, alongside existing models like ARIMA,
HoltWinters, and Timer-XL.
## Motivation
1. **Prophet** is a widely-used forecasting library that handles
seasonality, holidays, and missing data automatically. It's production-proven
at scale.
2. **StatsForecaster** provides efficient statistical models (AutoARIMA,
AutoETS, AutoTheta) that are significantly faster than traditional
implementations while maintaining accuracy.
3. Both libraries are already common dependencies in time-series forecasting
pipelines, making IoTDB a more complete solution for industrial IoT analytics.
## Proposed Changes
### 1. ConfigNode (Java)
Add models to the whitelist in `ModelInfo.java`:
```java
//
iotdb-core/confignode/src/main/java/org/apache/iotdb/confignode/persistence/ModelInfo.java
static {
builtInForecastModel.add("arima");
builtInForecastModel.add("naive_forecaster");
// ... existing models ...
builtInForecastModel.add("prophet"); // NEW
builtInForecastModel.add("statsforecast"); // NEW
}
```
### 2. AINode (Python)
**constant.py** - Add Prophet attributes:
```python
class AttributeName(Enum):
# Prophet attributes
YEARLY_SEASONALITY = "yearly_seasonality"
WEEKLY_SEASONALITY = "weekly_seasonality"
DAILY_SEASONALITY = "daily_seasonality"
SEASONALITY_MODE = "seasonality_mode"
```
**model_info.py** - Add model enum and mapping:
```python
class BuiltInModelType(Enum):
PROPHET = "Prophet"
STATSFORECASTER = "StatsForecaster"
BUILT_IN_MACHINE_LEARNING_MODEL_MAP = {
"prophet": ModelInfo(model_id="prophet", ...),
"statsforecast": ModelInfo(model_id="statsforecast", ...),
}
```
**built_in_model_factory.py** - Add model implementations.
### 3. Dependencies
Add to AINode requirements:
```
prophet>=1.1.5
statsforecast>=1.7.0
```
## Usage Example
```sql
-- Prophet forecast
SELECT prophet(temperature, 'predict_length'='24',
'daily_seasonality'='true')
FROM root.factory.sensor1
WHERE time >= 2024-01-01 AND time < 2024-02-01
-- StatsForecaster with AutoARIMA
SELECT statsforecast(temperature, 'predict_length'='24', 'model'='autoarima')
FROM root.factory.sensor1
WHERE time >= 2024-01-01 AND time < 2024-02-01
```
## Benefits
- **Performance**: Built-in models execute 7-20x faster than API-based
approaches due to eliminated Python client overhead
- **Simplicity**: Native SQL interface without requiring external Python code
- **Production-ready**: Both libraries are battle-tested in production
environments
## Additional Context
I have working prototype patches for the AINode side. Happy to contribute a
PR if there's interest.
/cc @apache/iotdb-committers
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
