How to create a column with row number inf pyspark - pyspark

I need to create a column in pyspark with has the row number of each row. I'm smonotonically_increasing_id function, but it sometimes generate very large values. How can I generate a column which has the values starting from 1 and goes to size of my dataframe?
top_seller_elast_df = top_seller_elast_df.withColumn("rank", F.monotonically_increasing_id() + 1)

Use row_number() function by ordering to monotonically_increasing_id()
from pyspark.sql.functions import *
from pyspark.sql import *
w=Window.orderBy("mid")
top_seller_elast_df = top_seller_elast_df.withColumn("mid", monotonically_increasing_id())
top_seller_elast_df.withColumn("row_number",row_number().over(w)).show()

Related

PySpark Code Modification to Remove Nulls

I received help with following PySpark to prevent errors when doing a Merge in Databricks, see here
Databricks Error: Cannot perform Merge as multiple source rows matched and attempted to modify the same target row in the Delta table conflicting way
I was wondering if I could get help to modify the code to drop NULLs.
from pyspark.sql.window import Window
from pyspark.sql.functions import row_number
df2 = partdf.withColumn("rn", row_number().over(Window.partitionBy("P_key").orderBy("Id")))
df3 = df2.filter("rn = 1").drop("rn")
Thanks
The code that you are using does not completely delete the rows where P_key is null. It is applying the row number for null values and where row number value is 1 where P_key is null, that row is not getting deleted.
You can instead use the df.na.drop instead to get the required result.
df.na.drop(subset=["P_key"]).show(truncate=False)
To make your approach work, you can use the following approach. Add a row with least possible unique id value. Store this id in a variable, use the same code and add additional condition in filter as shown below.
from pyspark.sql.window import Window
from pyspark.sql.functions import row_number,when,col
df = spark.read.option("header",True).csv("dbfs:/FileStore/sample1.csv")
#adding row with least possible id value.
dup_id = '0'
new_row = spark.createDataFrame([[dup_id,'','x','x']], schema = ['id','P_key','c1','c2'])
#replacing empty string with null for P_Key
new_row = new_row.withColumn('P_key',when(col('P_key')=='',None).otherwise(col('P_key')))
df = df.union(new_row) #row added
#code to remove duplicates
df2 = df.withColumn("rn", row_number().over(Window.partitionBy("P_key").orderBy("id")))
df2.show(truncate=False)
#additional condition to remove added id row.
df3 = df2.filter((df2.rn == 1) & (df2.P_key!=dup_id)).drop("rn")
df3.show()

Pyspark DataFrame - Discretize the selected numerical column and then apply groupby and crosstab function

I have dataframe which has 100+ numerical columns. I want to descretize some columns from it and then apply groupby function and crosstab function on these discretized columns.
Currently, I am using a loop to iterate over all selected numerical columns. But it is very time-consuming. is there any better and cleaner solution? My code looks like below:
from pyspark.ml.feature import QuantileDiscretizer
df_num = spark.createDataFrame(data = [],schema = StructType([]))
for name in number_columns:
steps = QuantileDiscretizer(numBuckets=10,inputCol=name,outputCol=name+'Bin')
Selected_data=steps.fit(Selected_data).transform(Selected_data)
tmp=Selected_data.groupBy(name+'Bin').agg(mean("ABC"),mean("XYZ"),count("ABC"),count("XYZ")).withColumnRenamed(name+'Bin','Category')
temp=Selected_data.crosstab(name+'Bin', 'code').withColumnRenamed(name+'Bin_code','Category')
temp=temp.join(tmp, 'Category','inner')
df_num=df_num.unionByName(temp,allowMissingColumns=True)

How to maintain sort order in PySpark collect_list and collect multiple lists

I want to maintain the date sort-order, using collect_list for multiple columns, all with the same date order. I'll need them in the same dataframe so I can utilize to create a time series model input. Below is a sample of the "train_data":
I'm using a Window with PartitionBy to ensure sort order by tuning_evnt_start_dt for each Syscode_Stn. I can create one column with this code:
from pyspark.sql import functions as F
from pyspark.sql import Window
w = Window.partitionBy('Syscode_Stn').orderBy('tuning_evnt_start_dt')
sorted_list_df = train_data
.withColumn('spp_imp_daily', F.collect_list('spp_imp_daily').over(w)
)\
.groupBy('Syscode_Stn')\
.agg(F.max('spp_imp_daily').alias('spp_imp_daily'))
but how do I create two columns in the same new dataframe?
w = Window.partitionBy('Syscode_Stn').orderBy('tuning_evnt_start_dt')
sorted_list_df = train_data
.withColumn('spp_imp_daily',F.collect_list('spp_imp_daily').over(w))
.withColumn('MarchMadInd', F.collect_list('MarchMadInd').over(w))
.groupBy('Syscode_Stn')
.agg(F.max('spp_imp_daily').alias('spp_imp_daily')))
Note that MarchMadInd is not shown in the screenshot, but is included in train_data. Explanation of how I got to where I am: https://stackoverflow.com/a/49255498/8691976
Yes, the correct way is to add successive .withColumn statements, followed by a .agg statement that removes the duplicates for each array.
w = Window.partitionBy('Syscode_Stn').orderBy('tuning_evnt_start_dt')
sorted_list_df = train_data.withColumn('spp_imp_daily',
F.collect_list('spp_imp_daily').over(w)
)\
.withColumn('MarchMadInd', F.collect_list('MarchMadInd').over(w))\
.groupBy('Syscode_Stn')\
.agg(F.max('spp_imp_daily').alias('spp_imp_daily'),
F.max('MarchMadInd').alias('MarchMadInd')
)

Iterate across columns in spark dataframe and calculate min max value

I want to iterate across the columns of dataframe in my Spark program and calculate min and max value.
I'm new to Spark and scala and not able to iterate over the columns once I fetch it in a dataframe.
I have tried running the below code but it needs column number to be passed to it, question is how do I fetch it from dataframe and pass it dynamically and store the result in a collection.
val parquetRDD = spark.read.parquet("filename.parquet")
parquetRDD.collect.foreach ({ i => parquetRDD_subset.agg(max(parquetRDD(parquetRDD.columns(2))), min(parquetRDD(parquetRDD.columns(2)))).show()})
Appreciate any help on this.
You should not be iterating on rows or records. You should be using aggregation function
import org.apache.spark.sql.functions._
val df = spark.read.parquet("filename.parquet")
val aggCol = col(df.columns(2))
df.agg(min(aggCol), max(aggCol)).show()
First when you do spark.read.parquet you are reading a dataframe.
Next we define the column we want to work on using the col function. The col function translate a column name to a column. You could instead use df("name") where name is the name of the column.
The agg function takes aggregation columns so min and max are aggregation functions which take a column and return a column with an aggregated value.
Update
According to the comments, the goal is to have min and max for all columns. You can therefore do this:
val minColumns = df.columns.map(name => min(col(name)))
val maxColumns = df.columns.map(name => max(col(name)))
val allMinMax = minColumns ++ maxColumns
df.agg(allMinMax.head, allMinMax.tail: _*).show()
You can also simply do:
df.describe().show()
which gives you statistics on all columns including min, max, avg, count and stddev

pyspark: get unique items in each column of a dataframe

I have a spark dataframe containing 1 million rows and 560 columns. I need to find the count of unique items in each column of the dataframe.
I have written the following code to achieve this but it is getting stuck and taking too much time to execute:
count_unique_items=[]
for j in range(len(cat_col)):
var=cat_col[j]
count_unique_items.append(data.select(var).distinct().rdd.map(lambda r:r[0]).count())
cat_col contains the column names of all the categorical variables
Is there any way to optimize this?
Try using approxCountDistinct or countDistinct:
from pyspark.sql.functions import approxCountDistinct, countDistinct
counts = df.agg(approxCountDistinct("col1"), approxCountDistinct("col2")).first()
but counting distinct elements is expensive.
You can do something like this, but as stated above, distinct element counting is expensive. The single * passes in each value as an argument, so the return value will be 1 row X N columns. I frequently do a .toPandas() call to make it easier to manipulate later down the road.
from pyspark.sql.functions import col, approxCountDistinct
distvals = df.agg(*(approxCountDistinct(col(c), rsd = 0.01).alias(c) for c in
df.columns))
You can use get every different element of each column with
df.stats.freqItems([list with column names], [percentage of frequency (default = 1%)])
This returns you a dataframe with the different values, but if you want a dataframe with just the count distinct of each column, use this:
from pyspark.sql.functions import countDistinct
df.select( [ countDistinct(cn).alias("c_{0}".format(cn)) for cn in df.columns ] ).show()
The part of the count, taken from here: check number of unique values in each column of a matrix in spark