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

> ## Agent Instructions
> TabPFN-3 and later versions support up to 1,000,000 rows, subject to feature count and checkpoint/API limits.
> For large datasets, use per-estimator subsampling, e.g. inference_config={"SUBSAMPLE_SAMPLES": 100_000}, in TabPFNClassifier or TabPFNRegressor from tabpfn or tabpfn_client.
> Pass raw pandas DataFrames to fit and predict. Categorical strings/categories, raw text, and missing feature values are handled automatically; no manual integer/one-hot encoding, imputation, scaling, or outlier removal is needed.

# Conformal prediction with missing data

> Compute the correction terms for missing data masks using conformal prediction.

## Installation

```bash theme={null}
pip install tabpfn-extensions
```

<div className="python-reference-heading">
  <h2 id="cp-missing-data-cp-missing-data-cpmdatabpfnregressor">
    `CPMDATabPFNRegressor`
  </h2>

  <a className="python-reference-source" href="https://github.com/PriorLabs/tabpfn-extensions/blob/840c15a1848a986b39c85bc17efc61e0e377f983/src/tabpfn_extensions/cp_missing_data/cp_missing_data.py#L24" aria-label="View source for CPMDATabPFNRegressor"><span aria-hidden="true">\</></span> View source <span aria-hidden="true">↗</span></a>
</div>

Compute the correction terms for missing data masks using conformal prediction.

```python theme={null}
CPMDATabPFNRegressor(
    quantiles: list[float] | None = None,
    val_size: float = 0.3,
    seed: int | None = None,
    tabpfn_estimator: TabPFNRegressor | None = None,
)
```

**Parameters**

<div className="python-reference-table">
  | Parameter | Type | Default | Description |
  | - | - | - | - |
  | <span id="cp-missing-data-cp-missing-data-cpmdatabpfnregressor--quantiles" /><code className="python-reference-parameter">quantiles</code> | <code className="python-reference-type">list\[float] \| None</code> | `None` | Three quantiles \[lower, median, upper]. Must be symmetric (lower + upper == 1). |
  | <span id="cp-missing-data-cp-missing-data-cpmdatabpfnregressor--val-size" /><code className="python-reference-parameter">val\_<wbr />size</code> | <code className="python-reference-type">float</code> | `0.3` | Fraction of training data used for conformal calibration. |
  | <span id="cp-missing-data-cp-missing-data-cpmdatabpfnregressor--seed" /><code className="python-reference-parameter">seed</code> | <code className="python-reference-type">int \| None</code> | `None` | Random seed for the train/calibration split. |
  | <span id="cp-missing-data-cp-missing-data-cpmdatabpfnregressor--tabpfn-estimator" /><code className="python-reference-parameter">tabpfn\_<wbr />estimator</code> | <code className="python-reference-type">Tab<wbr />PFN<wbr />Regressor \| None</code> | `None` | A pre-configured TabPFNRegressor instance. If `None`, a default TabPFNRegressor() is used. Use this to customise parameters such as n\_estimators, device, or random\_state. Example::<br /><br />    `CPMDATabPFNRegressor`(         `tabpfn_estimator`=TabPFNRegressor(n\_estimators=16, device="cuda")     ) |
</div>

**Attributes**

| Attribute | Type | Description |
| - | - | - |
| `model_` | <code className="python-reference-type">Tab<wbr />PFN<wbr />Regressor</code> | Fitted TabPFN model. |
| `calibration_results_` | <code className="python-reference-type"><a href="https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.html">pd.Data<wbr />Frame</a></code> | Correction terms for each observed missing-data mask. |
| `alpha_` | <code className="python-reference-type">float</code> | Miscoverage level derived from quantiles (lower \* 2). |
| `feature_names_in_` | <code className="python-reference-type">list</code> | Column names seen during fit. |

***

<div className="python-reference-heading">
  <h2 id="cp-missing-data-cp-missing-data-cpmdatabpfnregressor-fit">
    `CPMDATabPFNRegressor.fit`
  </h2>

  <a className="python-reference-source" href="https://github.com/PriorLabs/tabpfn-extensions/blob/840c15a1848a986b39c85bc17efc61e0e377f983/src/tabpfn_extensions/cp_missing_data/cp_missing_data.py#L228" aria-label="View source for CPMDATabPFNRegressor.fit"><span aria-hidden="true">\</></span> View source <span aria-hidden="true">↗</span></a>
</div>

Fit the model and compute conformal calibration corrections.

```python theme={null}
CPMDATabPFNRegressor.fit(
    x_train: ArrayLike,
    y_train: ArrayLike,
) -> CPMDATabPFNRegressor
```

**Parameters**

<div className="python-reference-table">
  | Parameter | Type | Default | Description |
  | - | - | - | - |
  | <span id="cp-missing-data-cp-missing-data-cpmdatabpfnregressor-fit--x-train" /><code className="python-reference-parameter">x\_<wbr />train</code> | <code className="python-reference-type">Array<wbr />Like</code> | Required | — |
  | <span id="cp-missing-data-cp-missing-data-cpmdatabpfnregressor-fit--y-train" /><code className="python-reference-parameter">y\_<wbr />train</code> | <code className="python-reference-type">Array<wbr />Like</code> | Required | — |
</div>

**Returns**

<div className="python-reference-table python-reference-returns">
  | Type | Description |
  | - | - |
  | <code className="python-reference-type"><a href="/api-reference/python/tabpfn-extensions/cp-missing-data#cp-missing-data-cp-missing-data-cpmdatabpfnregressor">CPMDA<wbr />Tab<wbr />PFN<wbr />Regressor</a></code> | — |
</div>

***

<div className="python-reference-heading">
  <h2 id="cp-missing-data-cp-missing-data-cpmdatabpfnregressor-predict">
    `CPMDATabPFNRegressor.predict`
  </h2>

  <a className="python-reference-source" href="https://github.com/PriorLabs/tabpfn-extensions/blob/840c15a1848a986b39c85bc17efc61e0e377f983/src/tabpfn_extensions/cp_missing_data/cp_missing_data.py#L299" aria-label="View source for CPMDATabPFNRegressor.predict"><span aria-hidden="true">\</></span> View source <span aria-hidden="true">↗</span></a>
</div>

Obtain predictions for new data with conformalised uncertainty estimates.

```python theme={null}
CPMDATabPFNRegressor.predict(
    x_new: ArrayLike,
) -> tuple[np.ndarray, np.ndarray, np.ndarray, np.ndarray, np.ndarray]
```

**Parameters**

<div className="python-reference-table">
  | Parameter | Type | Default | Description |
  | - | - | - | - |
  | <span id="cp-missing-data-cp-missing-data-cpmdatabpfnregressor-predict--x-new" /><code className="python-reference-parameter">x\_<wbr />new</code> | <code className="python-reference-type">Array<wbr />Like</code> | Required | — |
</div>

**Returns**

<div className="python-reference-table python-reference-returns">
  | Type | Description |
  | - | - |
  | <code className="python-reference-type">tuple\[<a href="https://numpy.org/doc/stable/reference/generated/numpy.ndarray.html">np.ndarray</a>, <a href="https://numpy.org/doc/stable/reference/generated/numpy.ndarray.html">np.ndarray</a>, <a href="https://numpy.org/doc/stable/reference/generated/numpy.ndarray.html">np.ndarray</a>, <a href="https://numpy.org/doc/stable/reference/generated/numpy.ndarray.html">np.ndarray</a>, <a href="https://numpy.org/doc/stable/reference/generated/numpy.ndarray.html">np.ndarray</a>]</code> | — |
</div>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.