Union Multiple dataframes in loop, with different schema

KKo
Contributor III

‎03-28-2022 12:47 PM

With in a loop I have few dataframes created. I can union them with out an issue if they have same schema using (df_unioned = reduce(DataFrame.unionAll, df_list). Now my problem is how to union them if one of the dataframe in df_list has different number of columns? I thought, reduce(df_unioned=DataFrame.unionByName, df_list, allowMissingColumns=True) would solve the issue but it is giving me error: reduce() takes no keyword arguments. Thanks in advance. Let me know if you need any details in the question.

0 Kudos

Anonymous
Not applicable

‎03-29-2022 05:00 AM

Union doesn't work if they have different schemas and columns. If you do need to union dataframes with different schemas, just add columns of nulls for anything missing to get them to the same schema.

KKo
Contributor III
In response to Anonymous

‎03-31-2022 08:19 AM

@Joseph Kambourakis​  I found a way to achieve this. using the function

def union_all(dfs):

 if len(dfs) > 1:

  return dfs[0].unionByName(union_all(dfs[1:]), allowMissingColumns=True)

 else:

  return dfs[0]

View solution in original post

anoopunni
New Contributor II

‎07-23-2023 08:47 PM - edited ‎07-23-2023 08:58 PM

Hi,
I have come across same scenario, using reduce() and unionByname we can implement the solution as below:

val lstDF: List[Datframe] = List(df1,df2,df3,df4,df5)

val combinedDF = lstDF.reduce((df1, df2) => df1.unionByName(df2, allowMissingColumns = true))

#Scala # Spark #multiple schema

0 Kudos