> ## Documentation Index
> Fetch the complete documentation index at: https://nixtlaverse.nixtla.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Exogenous Variables

> Exogenous variable configuration and generation

### `ExogenousConfig`

Bases: <code>[BaseModel](#pydantic.BaseModel)</code>

Configuration for exogenous variable generation.

Controls which exogenous columns are added to the output DataFrame.
All options are off by default for backward compatibility.

**Parameters:**

| Name                | Type | Description                                              | Default    |
| ------------------- | ---- | -------------------------------------------------------- | ---------- |
| `datetime_features` |      | Add calendar features (year, month, day\_of\_week, etc.) | *required* |
| `datetime_cyclical` |      | Add sin/cos cyclical encodings of datetime features      | *required* |
| `anomaly_flags`     |      | Add binary column indicating anomaly positions           | *required* |
| `changepoint_flags` |      | Add binary column indicating changepoint positions       | *required* |
| `missing_flags`     |      | Add binary column indicating missing data positions      | *required* |
| `correlated`        |      | List of correlated exogenous variables to generate       | *required* |

#### `ExogenousConfig.anomaly_flags`

```python theme={null}
anomaly_flags: bool = Field(default=False, description='Add anomaly indicator column')
```

#### `ExogenousConfig.changepoint_flags`

```python theme={null}
changepoint_flags: bool = Field(default=False, description='Add changepoint indicator column')
```

#### `ExogenousConfig.correlated`

```python theme={null}
correlated: list[CorrelatedExogConfig] = Field(default_factory=list, description='Correlated exogenous variables to generate')
```

#### `ExogenousConfig.datetime_cyclical`

```python theme={null}
datetime_cyclical: bool = Field(default=False, description='Add sin/cos cyclical encodings')
```

#### `ExogenousConfig.datetime_features`

```python theme={null}
datetime_features: bool = Field(default=False, description='Add calendar features')
```

#### `ExogenousConfig.missing_flags`

```python theme={null}
missing_flags: bool = Field(default=False, description='Add missing data indicator column')
```

#### `ExogenousConfig.model_config`

```python theme={null}
model_config = ConfigDict(extra='forbid')
```

#### `ExogenousConfig.validate_unique_names`

```python theme={null}
validate_unique_names()
```

Reject duplicate output columns before dataframe construction.

### `CorrelatedExogConfig`

Bases: <code>[BaseModel](#pydantic.BaseModel)</code>

Configuration for a single correlated exogenous variable.

#### `CorrelatedExogConfig.correlation`

```python theme={null}
correlation: float = Field(default=0.7, ge=(-1), le=1, description='Target correlation with the series')
```

#### `CorrelatedExogConfig.lag`

```python theme={null}
lag: int = Field(default=1, ge=1, description='Lag for lagged_copy method')
```

#### `CorrelatedExogConfig.method`

```python theme={null}
method: Literal['correlated_noise', 'lagged_copy', 'trend_following'] = Field(default='correlated_noise', description='Method for generating correlated exogenous')
```

#### `CorrelatedExogConfig.model_config`

```python theme={null}
model_config = ConfigDict(extra='forbid')
```

#### `CorrelatedExogConfig.name`

```python theme={null}
name: str = Field(..., description='Column name for this exogenous variable')
```

#### `CorrelatedExogConfig.noise_std`

```python theme={null}
noise_std: float = Field(default=0.1, ge=0, description='Noise std for lagged_copy method')
```

#### `CorrelatedExogConfig.smoothing_window`

```python theme={null}
smoothing_window: int = Field(default=10, ge=1, description='Window size for trend_following method')
```

#### `CorrelatedExogConfig.trend_noise_std`

```python theme={null}
trend_noise_std: float = Field(default=0.1, ge=0, description='Noise std for trend_following method')
```