如何在自定义的PySpark ML流水线_transform()方法中创建一个Spark DataFrame？

from pyspark.sql import SparkSession
from pyspark.sql import DataFrame

spark = SparkSession.builder.getOrCreate()

def _transform(self, dataset: DataFrame) -> DataFrame:
    # 创建自定义的DataFrame
    custom_df = spark.createDataFrame([(1, 'A'), (2, 'B'), (3, 'C')], ['id', 'name'])
    
    # 返回新的DataFrame
    return custom_df

# 假设pipeline是一个已定义的流水线对象
pipeline_model = pipeline.fit(input_data)
output_data = pipeline_model.transform(input_data)