@@ -17,13 +17,17 @@ public class TypedDataset {
1717public static void main (String []args )throws Exception {
1818
1919Logger .getLogger ("org" ).setLevel (Level .ERROR );
20- SparkSession session =SparkSession .builder ().appName ("StackOverFlowSurvey" ).master ("local[1 ]" ).getOrCreate ();
20+ SparkSession session =SparkSession .builder ().appName ("StackOverFlowSurvey" ).master ("local[* ]" ).getOrCreate ();
2121
2222DataFrameReader dataFrameReader =session .read ();
2323
2424Dataset <Row >responses =dataFrameReader .option ("header" ,"true" ).csv ("in/2016-stack-overflow-survey-responses.csv" );
2525
26- Dataset <Row >responseWithSelectedColumns =responses .select (col ("country" ),col ("age_midpoint" ).as ("ageMidPoint" ).cast ("integer" ),col ("occupation" ),col ("salary_midpoint" ).as ("salaryMidPoint" ).cast ("integer" ));
26+ Dataset <Row >responseWithSelectedColumns =responses .select (
27+ col ("country" ),
28+ col ("age_midpoint" ).as ("ageMidPoint" ).cast ("integer" ),
29+ col ("occupation" ),
30+ col ("salary_midpoint" ).as ("salaryMidPoint" ).cast ("integer" ));
2731
2832Dataset <Response >typedDataset =responseWithSelectedColumns .as (Encoders .bean (Response .class ));
2933
@@ -33,13 +37,13 @@ public static void main(String[] args) throws Exception {
3337System .out .println ("=== Print 20 records of responses table ===" );
3438typedDataset .show (20 );
3539
36- System .out .println ("=== Printrecords where theresponse is from Afghanistan ===" );
40+ System .out .println ("=== Print theresponses from Afghanistan ===" );
3741typedDataset .filter (response ->response .getCountry ().equals ("Afghanistan" )).show ();
3842
3943System .out .println ("=== Print the count of occupations ===" );
4044typedDataset .groupBy (typedDataset .col ("occupation" )).count ().show ();
4145
42- System .out .println ("=== Printrecords with average mid age less than 20 ===" );
46+ System .out .println ("=== Printresponses with average mid age less than 20 ===" );
4347typedDataset .filter (response ->response .getAgeMidPoint () !=null &&response .getAgeMidPoint () <20 ).show ();
4448
4549System .out .println ("=== Print the result with salary middle point in descending order ===" );
