Where and Filter in Spark Dataframes
Filtering rows from dataframe is one of the basic tasks performed when analyzing data with Spark. Spark provides two ways to filter data. Where and Filter function. Both of these functions work in the same way, but mostly we will be using “where” due to its familiarity with SQL.
Using Where / Filter in Spark Dataframe
We can easily filter rows with some conditions as we do in SQL using “Where” function. Say we need to find all rows where the number of flights is more than 50 between the two countries.
|
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 |
df_csv = spark.read.format("csv") \ .option("inferSchema", "true") \ .option("header","true") \ .load("data/flights.csv") df_csv.where("count > 50").show(5) +-----------------+-------------------+-----+ |DEST_COUNTRY_NAME|ORIGIN_COUNTRY_NAME|count| +-----------------+-------------------+-----+ | United States| Ireland| 344| | United States| India| 62| | United States| Grenada| 62| | Costa Rica| United States| 588| | United States| Sint Maarten| 325| +-----------------+-------------------+-----+ only showing top 5 rows |
We can also use column expressions. This time we will use “Filter” function to get desired rows from dataframe.
|
1 2 |
from pyspark.sql.functions import col df_csv.filter(col("DEST_COUNTRY_NAME") == "United States").show(5) |
Chaining Multiple Conditions
Though it is possible to write multiple where conditions in one statement, it is not necessary. Even when we chain multiple conditions one after another while creating a physical plan for execution spark will optimize these operations in one single step.
That is why it is always a better idea to write multiple where conditions separately which will be easier to understand while reading code.
|
1 2 3 |
df_csv.where("DEST_COUNTRY_NAME == 'United States'") \ .where("count>50") \ .show(5) |
I hope you found this useful :). See you in next blog.
Hi Mahesh ,
Could you pls share the Data set which you used in your example , it will be great help in practice
Hello Vivek,
I have committed this data at below git repo
https://github.com/mogalmahesh/spark-basics
I hope this helps.
Thanks