LagLeadTransformer<T>
Models & Types · AiDotNet.Preprocessing.TimeSeries
Creates lagged and leading features from time series data.
Inheritance: TimeSeriesTransformerBase<T> → LagLeadTransformer<T>
For Beginners
Lagged features capture historical information:
For example, if you want to predict today’s stock price:
- Lag 1: Yesterday’s price
- Lag 2: Price from 2 days ago
- Lag 7: Price from a week ago
Lead features capture future values (useful for targets, not features):
- Lead 1: Tomorrow’s price (what you might want to predict)
Why lag features matter:
- Many time series have autocorrelation (past values predict future)
- They help models learn temporal patterns
- They’re the simplest form of “memory” for a model
Example: Original: [100, 101, 102, 103, 104] Lag-1: [NaN, 100, 101, 102, 103] Lag-2: [NaN, NaN, 100, 101, 102]
How It Works
This transformer shifts data backward (lag) or forward (lead) in time to create features representing past or future values at each time step.
Constructors
| Constructor | Summary |
|---|---|
LagLeadTransformer(TimeSeriesFeatureOptions options) |
Creates a new lag/lead transformer with the specified options. |
Properties
| Property | Summary |
|---|---|
bool SupportsInverseTransform { get; } |