Pyspark : Problem with FloatType while writing data to parquet file

Question

I am having following schema,

root
 |-- A: string (nullable = true)
 |-- B: float (nullable = true)

And when I apply schema on data, dataframe va