Inner Join creating duplicates in SQL along with Group By

• lynhudson1

  Old Hand

  Points: 332

  More actions

  January 20, 2021 at 10:00 am

  #3835422

  I have two datasets, where I would like to join and then group by a particular column

  df1

  version host date      name purpose
  pat     a16  12/1/2019 pat  hi
  fam     a16  12/1/2019 fam  hi
  emp     a16  12/1/2019 emp  hi
  dan     a16  12/1/2019 dan  hi
  

  df2

  pc  size free date
  a16 5    1    12/1/2019
  a40 4    3    12/1/2019
  a41 3    1    12/1/2019

  Desired

  host date      purpose pc  size free
  a16  12/1/2019 hi      a16 5    1

  Doing

  SELECT 
  sum(size) AS size, sum(free) AS free,
  df1.purpose, df1.date, df1.host
  FROM df1
  JOIN df2 ON
  df1.host = df2.pc AND
  df1.date = df2.date
  GROUP BY df1.purpose, df1.date, df1.host

  The problem is that this join keeps giving me this below with the sum of size being

  size 20 and free 4, when in actuality, the sum of size should be 5 and free should be 1.

  I am trying to reflect the above in a query. Any suggestion is appreciated.

   

  
  host date      purpose pc size free
  a16  12/1/2019 hi      a16 5   1
  a16  12/1/2019 hi      a16 5   1
  a16  12/1/2019 hi      a16 5   1
  a16  12/1/2019 hi      a16 5   1

• Ant-Green

  SSC Guru

  Points: 113545

  More actions

  January 20, 2021 at 10:49 am

  #3835448

  Thats down to only joining df1 and df2 on the host/pc column.  You have 4 different versions/names in df1.

  What was to happen should the purpose change to "bye" for example, how is that supposed to be calculated.

  It sounds like your missing a crucial piece of information in df2 to be able to join the two pieces of information together

  But you would need to do is to group the values from df1 first before joining them to df2, something like the below

  SELECT 
  sum(size) AS size, sum(free) AS free,
  df1.purpose, df1.date, df1.host
  FROM (SELECT host, date, purpose FROM df1 GROUP BY host, date, purpose) as groupedDF1
  JOIN df2 ON
  groupeddf1.host = df2.pc AND
  groupeddf1.date = df2.date
  GROUP BY groupeddf1.purpose, groupeddf1.date, groupeddf1.host