Pyspark get max value exclude NaN

Question

df = spark.createDataFrame([(1.0,100.0, float('nan')), (float('nan'),100.0, 2.0)], ("a", "b"))
df.select(F.max(df.a)).show()

the result is :

+------+
|max(a)|
+------+
|   NaN|
+------+

I didn't want the NaN,how can I do?

Answer 1

you can do

df.na.fill(0) 

before you apply your max filter