init research

dataframe/docs/StardustDocs/topics/cast.md

@@ -0,0 +1,64 @@
+[//]: # (title: cast)
+<!---IMPORT org.jetbrains.kotlinx.dataframe.samples.api.Modify-->
+
+Changes the type argument of the [`DataFrame`](DataFrame.md) instance without changing its contents.
+
+```kotlin
+cast<T>(verify = false)
+```
+
+Related operations: [](adjustSchema.md)
+
+**Parameters:**
+* `verify: Boolean = false` —
+  when `true`, the function throws an exception if the [`DataFrame`](DataFrame.md) instance doesn't match the given schema. 
+Otherwise, it just changes the format type without actual data checks.
+
+Use this operation to change the formal type of a [`DataFrame`](DataFrame.md) instance
+to match the expected schema and enable generated [extension properties](extensionPropertiesApi.md) for it.
+
+```kotlin
+@DataSchema
+interface Person {
+    val age: Int
+    val name: String
+}
+
+df.cast<Person>()
+```
+
+To convert [`DataFrame`](DataFrame.md) columns to match given schema, use [`convertTo`](convertTo.md) operation.
+
+**Reusing implicitly generated schema**
+
+```kotlin
+castTo<T>(df: DataFrame<T>)
+```
+
+In notebooks, dataframe types are implicitly generated.
+
+![Implicitly generated schema](implicitlyGeneratedSchema.png)
+
+This type can be referred to, but its name will change whenever you re-execute cells.
+Here how you can do it in a more robust way:
+
+<!---FUN castToGenerateSchema-->
+
+```kotlin
+val sample = DataFrame.readJson("sample.json")
+```
+
+<!---END-->
+
+<!---FUN castTo-->
+
+```kotlin
+for (file in files) {
+    // df here is expected to have the same structure as sample
+    val df = DataFrame.readJson(file).castTo(sample)
+    val count = df.count { perf > 10.0 }
+    println("$file: $count")
+}
+```
+
+<!---END-->