CODEWITHSUNDEEP
pythonapache-sparkpyspark
Jay
Jay

Reputation: 326

how to drop duplicates but keep first in pyspark dataframe?

I am trying to remove duplicates from data-frame but first entry should not be removed . excluding first record rest all other duplicates should get stored in one separate data-frame .

for e.g if data-frame is like :

col1,col2,col3,col4
r,t,s,t
a,b,c,d
b,m,c,d
a,b,c,d
a,b,c,d
g,n,d,f
e,f,g,h
t,y,u,o
e,f,g,h
e,f,g,h

in such case I should have two data-frames .

df1:
r,t,s,t
a,b,c,d
b,m,c,d
g,n,d,f
e,f,g,h
t,y,u,o

and other data-frame should be :

a,b,c,d
a,b,c,d
e,f,g,h
e,f,g,h

Upvotes: 2

Views: 7342

Answers (1)

notNull
notNull

Reputation: 31460

Try using window row_number() function.

Example:

df.show()
#+----+----+----+----+
#|col1|col2|col3|col4|
#+----+----+----+----+
#|   r|   t|   s|   t|
#|   a|   b|   c|   d|
#|   b|   m|   c|   d|
#|   a|   b|   c|   d|
#|   a|   b|   c|   d|
#|   g|   n|   d|   f|
#|   e|   f|   g|   h|
#|   t|   y|   u|   o|
#|   e|   f|   g|   h|
#|   e|   f|   g|   h|
#+----+----+----+----+

from pyspark.sql import *
from pyspark.sql.functions import *

w=Window.partitionBy("col1","col2","col3","col4").orderBy(lit(1))


df1=df.withColumn("rn",row_number().over(w)).filter(col("rn")==1).drop("rn")

df1.show()
#+----+----+----+----+
#|col1|col2|col3|col4|
#+----+----+----+----+
#|   b|   m|   c|   d|
#|   r|   t|   s|   t|
#|   g|   n|   d|   f|
#|   t|   y|   u|   o|
#|   a|   b|   c|   d|
#|   e|   f|   g|   h|
#+----+----+----+----+
df2=df.withColumn("rn",row_number().over(w)).filter(col("rn")>1).drop("rn")
df2.show()
#+----+----+----+----+
#|col1|col2|col3|col4|
#+----+----+----+----+
#|   a|   b|   c|   d|
#|   a|   b|   c|   d|
#|   e|   f|   g|   h|
#|   e|   f|   g|   h|
#+----+----+----+----+

Upvotes: 4

Related Questions