根据相同的列名加入多个 Pyspark 数据框

Question

I am new to Pyspark so that is why I am stuck with the following:

I have 5 dataframes and each dataframes has the same Primary Key called concern_code. I need to outer join all this dataframes together and need to drop the 4 columns called concern_code from the 4 dataframes.

For example: Dataframe Df1 outer joins Df2 based on concern_code Dataframe Df1 outer joins Df3 based on concern_code and so on..

My current Pyspark syntax looks like this:

df1.join(df2,["concern_code"])\
  .join(df3,df1["concern_code"] == df3["concern_code"])\
  .join(df4,df1["concern_code"] == df4["concern_code"])\
  .join(df5,df1["concern_code"] == df5["concern_code"])\
   .show()

How do I need to fix the syntax to perform outer join and to have a final version of a new dataframe that has only one column of concern_code ?

Answer 1

You are close. let's say you have following dfs:

d = [
    ("a", 5.2),
    ("b", 10.4),
    ("c", 7.8),
    ("d", 11.2),
    
]
df1 = spark.createDataFrame(d, ['concern_code','value'])

df2 = spark.createDataFrame(d, ['concern_code','value1'])

df3 = spark.createDataFrame(d, ['concern_code','value2'])

df4 = spark.createDataFrame(d, ['concern_code','value3'])

df5 = spark.createDataFrame(d, ['concern_code','value4'])

df1.show()

# output
+------------+-----+
|concern_code|value|
+------------+-----+
|           a|  5.2|
|           b| 10.4|
|           c|  7.8|
|           d| 11.2|
+------------+-----+

(
    df1
    .join(df2,on="concern_code", how="outer")
    .join(df3,on="concern_code", how="outer")
    .join(df4,on="concern_code", how="outer")
    .join(df5,on="concern_code", how="outer")
   .show()
)

# output
+------------+-----+------+------+------+------+
|concern_code|value|value1|value2|value3|value4|
+------------+-----+------+------+------+------+
|           c|  7.8|   7.8|   7.8|   7.8|   7.8|
|           d| 11.2|  11.2|  11.2|  11.2|  11.2|
|           a|  5.2|   5.2|   5.2|   5.2|   5.2|
|           b| 10.4|  10.4|  10.4|  10.4|  10.4|
+------------+-----+------+------+------+------+

Answer 2

If you join two data frames on columns then the columns will be duplicated, as in your case. So I would suggest to use an array of strings, or just a string, ie 'id', for joining two or more data frames.

The code below should not duplicate the column names:

df1.join(df2,on='id', how='outer')\
   .join(df3,on='id', how='outer')\
   .join(df4,on='id', how='outer')\
   .join(df5,on='id'  how='outer')\
   .show()