Average Line in 2013 NFL Games

AverageLine13.pig

-- analyze data from http://www.repole.com/sun4cast/data.html
raw_records = LOAD '/data/nfl/nfl2013lines.csv' USING PigStorage(',');
records = STREAM raw_records THROUGH `tail -n +2` AS
        (Date, Visitor, VisitorScore:int,
         HomeTeam, HomeScore:int, Line:float, TotalLine:float);
all_games = GROUP records ALL;
data = FOREACH all_games
  GENERATE AVG(records.VisitorScore) as Visitor, AVG(records.HomeScore) as Home, AVG(records.Line) as Spread;
data2 = FOREACH data
  GENERATE data.Home - data.Visitor as diff, data.Visitor, data.Home, data.Spread;
DUMP data2;
--STORE data2 INTO '/output/nfl/avg_line';