RollingStatsTransformer<T>
Models & Types ยท AiDotNet.Preprocessing.TimeSeries
Computes rolling statistics over time series data for feature engineering.
Inheritance: TimeSeriesTransformerBase<T> โ RollingStatsTransformer<T>
For Beginners
Rolling statistics are like taking a moving snapshot of your data.
For example, with a 7-day rolling mean:
- Day 7: Average of days 1-7
- Day 8: Average of days 2-8
- Day 9: Average of days 3-9
This helps capture trends and patterns at different time scales. Common uses include:
- Smoothing noisy data (rolling mean)
- Detecting volatility changes (rolling std deviation)
- Finding extreme values (rolling min/max)
- Understanding distribution changes (rolling skewness/kurtosis)
How It Works
This transformer calculates various statistical measures over rolling windows, creating features that capture local patterns in time series data.
Constructors
| Constructor | Summary |
|---|---|
RollingStatsTransformer(TimeSeriesFeatureOptions options) |
Creates a new rolling statistics transformer with the specified options. |
Properties
| Property | Summary |
|---|---|
bool SupportsInverseTransform { get; } |