init research

dataframe/docs/StardustDocs/topics/corr.md

@@ -0,0 +1,34 @@
+[//]: # (title: corr)
+
+Returns [`DataFrame`](DataFrame.md) with the pairwise correlation between two sets of columns.
+
+It computes the [Pearson correlation coefficient](https://en.wikipedia.org/wiki/Pearson_correlation_coefficient).
+
+```kotlin
+corr { columns1 }
+    .with { columns2 } | .withItself()
+```
+
+See [column selectors](ColumnSelectors.md) for how to select the columns for this operation.
+
+To compute pairwise correlation between all columns in the [`DataFrame`](DataFrame.md) use `corr` without arguments:
+
+```kotlin
+corr()
+```
+
+The function is available for numeric- and `Boolean` columns.
+`Boolean` values are converted into `1` for `true` and `0` for `false`.
+All other columns are ignored.
+
+If a [`ColumnGroup`](DataColumn.md#columngroup) instance is passed as the target column for correlation,
+it will be unpacked into suitable nested columns.
+
+The resulting [`DataFrame`](DataFrame.md) will have `n1` rows and `n2+1` columns,
+where `n1` and `n2` are the number of columns in `columns1` and `columns2` correspondingly.
+
+The first column will have the name "column" and will contain names of columns in `column1`.
+Other columns will have the same names as in `columns2` and will contain the computed correlation coefficients.
+
+If exactly one [`ColumnGroup`](DataColumn.md#columngroup) is passed in `columns1`,
+the first column in the output will have its name.