Allow for grouping of team formations

src/pages/9_Team_Formations.py

@@ -1,4 +1,5 @@
 import datetime
+import pandas as pd
 import streamlit as st
 
 from config import DEFAULT_ICON
@@ -10,7 +11,6 @@ from queries.nflverse.github_data import get_pbp_participation, get_current_tabl
 
 def load_data():
     data = get_pbp_participation(YEAR)
-    # data = data[data.fantasy_position]
     teams_list = sorted(filter(None, data.possession_team.unique()))
     # position_list = data.position.unique()
     # weeks_list = sorted(data.week.unique())
@@ -30,8 +30,9 @@ def get_page():
     data, teams_list, data_load_time_str = load_data()
     st.write(f"Data loaded as of: {data_load_time_str} UTC")
     default_groups = [
-        "possession_team",
-        "offense_formation",
+        "down",
+        "play_type",
+        "offense_personnel",
     ]
     group_options = [
         "week",
@@ -49,34 +50,32 @@ def get_page():
         "defense_personnel",
     ]
     group_by_selected = st.multiselect("Group by:", group_options) or default_groups
-    teams_selected = st.multiselect("Team:", teams_list, placeholder="Select a team to filter") or teams_list
+    team_selected = st.selectbox("Team:", teams_list)
+    week_selection = st.slider(
+        "Filter Week Range:",
+        min_value=data["week"].min(),
+        max_value=data["week"].max(),
+        value=(data["week"].min(), data["week"].max()),
+        step=1,
+    )
 
     with st.container():
-        filtered_data = data[(data.possession_team.isin(teams_selected))]
-        grouped_data = filtered_data.groupby(group_by_selected).size()
+        filtered_data = data[
+            (data.possession_team == team_selected)
+            & (data.play_type.isin(["pass", "run"]))
+            & (data["week"].between(*week_selection))
+        ]
+
         st.dataframe(
-            grouped_data,
-            hide_index=False,
-            # height=35 * (len(filtered_data) + 1) + 12,
+            pd.pivot_table(
+                filtered_data,
+                values="count_col",
+                index=group_by_selected,
+                columns="week",
+                aggfunc={"count_col": "sum"},
+                # margins=True,
+            ),
             use_container_width=False,
-            # column_order=[
-            #     "season",
-            #     "game_type",
-            #     "week",
-            #     "player",
-            #     "position",
-            #     "team",
-            #     "opponent",
-            #     "offense_snaps",
-            #     "offense_pct",
-            #     "defense_snaps",
-            #     "defense_pct",
-            #     "st_snaps",
-            #     "st_pct",
-            # ],
-            column_config={
-                "0": st.column_config.NumberColumn(label="Count"),
-            },
         )
 
 

src/queries/nflverse/github_data.py

@@ -52,10 +52,12 @@ def get_pbp_participation(season_int: int) -> pd.DataFrame:
                         , b.play_type
                         , b.pass_length
                         , b.pass_location
+                        , 1 as count_col
                     from pbp_participation_pbp_participation_{season_int} a
                     left join pbp_play_by_play_{season_int} b
                     on a.play_id = b.play_id
                     and a.nflverse_game_id = b.game_id
+                    where b.week is not null
 """
     ).df()
     return df