如何在PySpark中使用pandas轴等价来删除列而不是行？

Question

我有一个spark dataframe，如下所示

df = pd.DataFrame({
'subject_id':[1,1,1,1,2,2,2,2,3,3,4,4,4,4,4],
'readings' : ['READ_1','READ_2','READ_1','READ_3',np.nan,'READ_5',np.nan,'READ_8','READ_10','READ_12','READ_11','READ_14','READ_09','READ_08','READ_07'],
 'val' :[5,np.nan,7,np.nan,np.nan,7,np.nan,np.nan,np.nan,np.nan,np.nan,np.nan,np.nan,np.nan,46],
 })

from pyspark.sql.types import *
from pyspark.sql.functions import isnan, when, count, col

mySchema = StructType([ StructField("subject_id", LongType(), True)\
                       ,StructField("readings", StringType(), True)\
                       ,StructField("val", FloatType(), True)])

spark_df = spark.createDataFrame(df,schema=mySchema)

我想做的是删除包含超过80%的nan、NULL或0值的列？

我试过下面的东西，但不起作用

spark_df = spark_df.dropna(axis = 'columns',how=any,thresh=12)

Answer 1

可以使用Dropna方法中的subset参数指定要在其中查找空值的列。

若要移除null值超过80%的所有列：

columns_to_drop = []
count_before = spark_df.count()

for column_name in spark_df.columns:
    temp_spark_df =  spark_df.dropna(subset=[column_name], how=any, thresh=12)
    count_after = temp_spark_df.count()

    if ((count_before-count_after)/count_before) > 0.8:
        columns_to_drop.append(column_name)


spark_df = spark_df.drop(*columns_to_drop)