Address reviewer feedback: fix grammar, path resolution, and sampling logic

README.md

@@ -1,6 +1,6 @@
 # NBA Fantasy App
 
-This repository has a Streamlit app that lets you create a NBA fantasy team and compete against computer-generated teams. The app uses a local CSV file that contains historical NBA player data.
+This repository has a Streamlit app that lets you create an NBA fantasy team and compete against computer-generated teams. The app uses a local CSV file that contains historical NBA player data.
 
 To determine the winner of each game, the app uses a functional Keras model that was trained using the 2018 season's game by game results. The model uses season averages in stats from the starting five players as features against thier wins and losses. The training data and model script are included. 
 

src/database/connection.py

@@ -10,7 +10,8 @@ import streamlit as st
 
 logger = logging.getLogger("streamlit_nba")
 
-CSV_PATH = Path("snowflake_nba.csv")
+# Resolve path relative to this module
+CSV_PATH = Path(__file__).resolve().parent.parent.parent / "snowflake_nba.csv"
 
 
 class DatabaseConnectionError(Exception):
@@ -46,7 +47,8 @@ def load_data() -> pd.DataFrame:
         return df
     except Exception as e:
         logger.error(f"Failed to load CSV data: {e}")
-        raise DatabaseConnectionError(f"Could not load data from {CSV_PATH}: {e}") from e
+        msg = f"Could not load data from {CSV_PATH}: {e}"
+        raise DatabaseConnectionError(msg) from e
 
 
 @contextmanager

src/database/queries.py

@@ -91,16 +91,30 @@ def get_away_team_by_stats(
         DataFrame with 5 players
 
     Raises:
-        RuntimeError: If unable to get 5 players within max_attempts
+        QueryExecutionError: If unable to get 5 players within max_attempts
     """
     for attempt in range(max_attempts):
         try:
-            df1 = df[df["PTS"] > pts_threshold].sample(n=2)
-            df2 = df[df["REB"] > reb_threshold].sample(n=1)
-            df3 = df[df["AST"] > ast_threshold].sample(n=1)
-            df4 = df[df["STL"] > stl_threshold].sample(n=1)
-
-            results = pd.concat([df1, df2, df3, df4]).drop_duplicates()
+            # Sample without replacement across all picks
+            df1 = df[df["PTS"] > pts_threshold].sample(n=2, replace=False)
+
+            # For subsequent picks, exclude already chosen players
+            chosen = df1
+
+            df2 = df[
+                (df["REB"] > reb_threshold) & (~df.index.isin(chosen.index))
+            ].sample(n=1, replace=False)
+            chosen = pd.concat([chosen, df2])
+
+            df3 = df[
+                (df["AST"] > ast_threshold) & (~df.index.isin(chosen.index))
+            ].sample(n=1, replace=False)
+            chosen = pd.concat([chosen, df3])
+
+            df4 = df[
+                (df["STL"] > stl_threshold) & (~df.index.isin(chosen.index))
+            ].sample(n=1, replace=False)
+            results = pd.concat([chosen, df4])
 
             if len(results) == 5:
                 logger.info(f"Got away team on attempt {attempt + 1}")

tests/test_database.py

@@ -54,7 +54,9 @@ class TestGetPlayersByFullNames:
         assert set(result["FULL_NAME"]) == set(names)
         assert list(result.columns) == PLAYER_COLUMNS
 
-    def test_empty_names_returns_empty_dataframe(self, sample_player_df: pd.DataFrame) -> None:
+    def test_empty_names_returns_empty_dataframe(
+        self, sample_player_df: pd.DataFrame
+    ) -> None:
         """Test that empty input returns empty DataFrame."""
         result = get_players_by_full_names(sample_player_df, [])