# `SparkEx.DataFrame.Stat`
[🔗](https://github.com/lukaszsamson/spark_ex/blob/v0.1.1/lib/spark_ex/data_frame/stat.ex#L1)

Statistical operations sub-API for DataFrames.

Provides descriptive statistics, correlation, covariance, crosstab,
frequency items, approximate quantiles, and stratified sampling.

Most methods return lazy DataFrames. Scalar-returning methods
(`corr/4`, `cov/3`, `approx_quantile/4`) execute eagerly.

# `approx_quantile`

```elixir
@spec approx_quantile(
  SparkEx.DataFrame.t(),
  String.t() | [String.t()],
  [float()],
  float()
) :: {:ok, [float()] | [[float()]]} | {:error, term()}
```

Computes approximate quantiles for one or more columns.

Returns `{:ok, [float]}` for a single column or `{:ok, [[float]]}` for multiple.

## Examples

    {:ok, quantiles} = DataFrame.Stat.approx_quantile(df, "age", [0.25, 0.5, 0.75])
    {:ok, quantiles} = DataFrame.Stat.approx_quantile(df, ["age", "salary"], [0.5], 0.01)

# `corr`

```elixir
@spec corr(SparkEx.DataFrame.t(), String.t(), String.t(), String.t()) ::
  {:ok, float()} | {:error, term()}
```

Computes the Pearson correlation coefficient between two columns.

Returns `{:ok, float}` or `{:error, reason}`.

## Examples

    {:ok, r} = DataFrame.Stat.corr(df, "height", "weight")

# `cov`

```elixir
@spec cov(SparkEx.DataFrame.t(), String.t(), String.t()) ::
  {:ok, float()} | {:error, term()}
```

Computes the sample covariance between two columns.

Returns `{:ok, float}` or `{:error, reason}`.

## Examples

    {:ok, c} = DataFrame.Stat.cov(df, "height", "weight")

# `crosstab`

```elixir
@spec crosstab(SparkEx.DataFrame.t(), String.t(), String.t()) :: SparkEx.DataFrame.t()
```

Computes a contingency table (crosstab) of two columns.

Returns a DataFrame with the frequency of each combination of values.

## Examples

    DataFrame.Stat.crosstab(df, "department", "gender")

# `describe`

```elixir
@spec describe(SparkEx.DataFrame.t(), String.t() | [String.t()]) ::
  SparkEx.DataFrame.t()
```

Computes basic statistics (count, mean, stddev, min, max) for selected columns.

If no columns are given, describes all columns.

## Examples

    DataFrame.Stat.describe(df)
    DataFrame.Stat.describe(df, ["age", "salary"])

# `freq_items`

```elixir
@spec freq_items(SparkEx.DataFrame.t(), [String.t()], float() | keyword()) ::
  SparkEx.DataFrame.t()
```

Finds all items which have a frequency greater than or equal to `support`.

## Examples

    DataFrame.Stat.freq_items(df, ["category", "status"])
    DataFrame.Stat.freq_items(df, ["category"], 0.05)

# `sample_by`

```elixir
@spec sample_by(
  SparkEx.DataFrame.t(),
  SparkEx.Column.t() | String.t(),
  map(),
  integer() | keyword() | nil
) :: SparkEx.DataFrame.t()
```

Returns a stratified sample of the DataFrame.

## Parameters

- `col` — column name (string) or `Column` used for stratification.
- `fractions` — map of `%{stratum_value => sampling_fraction}`.
- `seed` — optional random seed. Auto-generated if not provided; pass an explicit seed for reproducibility.

## Examples

    DataFrame.Stat.sample_by(df, "label", %{0 => 0.1, 1 => 0.5})
    DataFrame.Stat.sample_by(df, "label", %{0 => 0.1, 1 => 0.5}, 42)

# `summary`

```elixir
@spec summary(SparkEx.DataFrame.t(), String.t() | [String.t()]) ::
  SparkEx.DataFrame.t()
```

Computes specified statistics for numeric and string columns.

Statistics can include: "count", "mean", "stddev", "min", "max",
and percentiles like "25%", "50%", "75%".

## Examples

    DataFrame.Stat.summary(df)
    DataFrame.Stat.summary(df, ["count", "min", "max"])

---

*Consult [api-reference.md](api-reference.md) for complete listing*