Преглед на файлове

Use good naming conventions

theenglishway (time) преди 2 години
родител
ревизия
b2353fb3fc
променени са 2 файла, в които са добавени 15 реда и са изтрити 15 реда
  1. 3 3
      src/main/scala/Main.scala
  2. 12 12
      src/main/scala/Transform.scala

+ 3 - 3
src/main/scala/Main.scala

@@ -32,9 +32,9 @@ object Main extends App {
     GamesAnalysis(localFiles("teams"), localFiles("games"), localFiles("stats"))
 
   Map(
-    "output/teams_games_df" -> analysis.teams_games.sort("game_id", "team_id"),
-    "output/stats_games_pts_df" -> analysis.stats_games_pts
+    "output/teams_games_df" -> analysis.teamsGamesDf.sort("game_id", "team_id"),
+    "output/stats_by_game_and_team" -> analysis.statsByGameAndTeamDf
       .sort("game_id", "team_id"),
-    "output/final_df" -> analysis.merged.sort("game_id", "team_id")
+    "output/final_df" -> analysis.finalDf.sort("game_id", "team_id")
   ).foreachEntry((path, df) => GamesAnalysis.writeToCsv(df, path))
 }

+ 12 - 12
src/main/scala/Transform.scala

@@ -11,13 +11,13 @@ case class GamesAnalysis(teams: os.Path, games: os.Path, stats: os.Path) {
     .getOrCreate()
   import spark.implicits._
 
-  val teams_df = readInput(teams)
-  val games_df = readInput(games)
-  val stats_df = readInput(stats)
+  val teamsDf = readInput(teams)
+  val gamesDf = readInput(games)
+  val statsDf = readInput(stats)
 
-  val teams_games = teams_df
+  val teamsGamesDf = teamsDf
     .as("t")
-    .join(games_df.as("g"))
+    .join(gamesDf.as("g"))
     .where(
       $"g.home_team.id" === $"t.id"
         || $"g.visitor_team.id" === $"t.id"
@@ -37,12 +37,12 @@ case class GamesAnalysis(teams: os.Path, games: os.Path, stats: os.Path) {
       $"team_score"
     )
 
-  val stats_games =
-    stats_df
-      .join(games_df, stats_df("game_id") === games_df("id"))
-      .join(teams_df, stats_df("team.id") === teams_df("id"))
+  val statsGamesDf =
+    statsDf
+      .join(gamesDf, statsDf("game_id") === gamesDf("id"))
+      .join(teamsDf, statsDf("team.id") === teamsDf("id"))
 
-  val stats_games_pts = stats_games
+  val statsByGameAndTeamDf = statsGamesDf
     .groupBy($"game.id".alias("game_id"), $"team.id".alias("team_id"))
     .agg(
       sum($"pts").alias("pts"),
@@ -51,9 +51,9 @@ case class GamesAnalysis(teams: os.Path, games: os.Path, stats: os.Path) {
       sum($"reb").alias("reb")
     )
 
-  val merged = teams_games
+  val finalDf = teamsGamesDf
     .as("tg")
-    .join(stats_games_pts.as("sgp"))
+    .join(statsByGameAndTeamDf.as("sgp"))
     .where(
       $"tg.game_id" === $"sgp.game_id"
         && $"tg.team_id" === $"sgp.team_id"