Outputs to slots

.gitignore

@@ -99,3 +99,4 @@ ENV/
 
 # mypy
 .mypy_cache/
+*.xlsx

data/NCAATourneySeeds.csv

@@ -2159,14 +2159,12 @@ Season,Seed,TeamID
 2018,W08,1439
 2018,W09,1104
 2018,W10,1139
-2018,W11a,1382
-2018,W11b,1417
+2018,W11,1382
 2018,W12,1293
 2018,W13,1267
 2018,W14,1372
 2018,W15,1168
-2018,W16a,1254
-2018,W16b,1347
+2018,W16,1347
 2018,X01,1242
 2018,X02,1181
 2018,X03,1277
@@ -2177,8 +2175,7 @@ Season,Seed,TeamID
 2018,X08,1371
 2018,X09,1301
 2018,X10,1328
-2018,X11a,1113
-2018,X11b,1393
+2018,X11,1393
 2018,X12,1308
 2018,X13,1158
 2018,X14,1137
@@ -2215,5 +2212,4 @@ Season,Seed,TeamID
 2018,Z13,1422
 2018,Z14,1285
 2018,Z15,1252
-2018,Z16a,1300
-2018,Z16b,1411
+2018,Z16,1411

data/NCAATourneySlots_Detailed_2018.csv

@@ -0,0 +1,65 @@
+Season,Slot,NextSlot,NextSeed,StrongSeed,WeakSeed,StrongTeamID,WeakTeamID,StrongTeamName,WeakTeamName
+2018,R1W1,R2W1,Strong,W01,W16,1437,1347,Villanova,Radford
+2018,R1W2,R2W2,Strong,W02,W15,1345,1168,Purdue,CS Fullerton
+2018,R1W3,R2W3,Strong,W03,W14,1403,1372,Texas Tech,SF Austin
+2018,R1W4,R2W4,Strong,W04,W13,1455,1267,Wichita St,Marshall
+2018,R1W5,R2W4,Weak,W05,W12,1452,1293,West Virginia,Murray St
+2018,R1W6,R2W3,Weak,W06,W11,1196,1382,Florida,St Bonaventure
+2018,R1W7,R2W2,Weak,W07,W10,1116,1139,Arkansas,Butler
+2018,R1W8,R2W1,Weak,W08,W09,1439,1104,Virginia Tech,Alabama
+2018,R1X1,R2X1,Strong,X01,X16,1242,1335,Kansas,Penn
+2018,R1X2,R2X2,Strong,X02,X15,1181,1233,Duke,Iona
+2018,R1X3,R2X3,Strong,X03,X14,1277,1137,Michigan St,Bucknell
+2018,R1X4,R2X4,Strong,X04,X13,1120,1158,Auburn,Col Charleston
+2018,R1X5,R2X4,Weak,X05,X12,1155,1308,Clemson,New Mexico St
+2018,R1X6,R2X3,Weak,X06,X11,1395,1393,TCU,Syracuse
+2018,R1X7,R2X2,Weak,X07,X10,1348,1328,Rhode Island,Oklahoma
+2018,R1X8,R2X1,Weak,X08,X09,1371,1301,Seton Hall,NC State
+2018,R1Y1,R2Y1,Strong,Y01,Y16,1438,1420,Virginia,UMBC
+2018,R1Y2,R2Y2,Strong,Y02,Y15,1153,1209,Cincinnati,Georgia St
+2018,R1Y3,R2Y3,Strong,Y03,Y14,1397,1460,Tennessee,Wright St
+2018,R1Y4,R2Y4,Strong,Y04,Y13,1112,1138,Arizona,Buffalo
+2018,R1Y5,R2Y4,Weak,Y05,Y12,1246,1172,Kentucky,Davidson
+2018,R1Y6,R2Y3,Weak,Y06,Y11,1274,1260,Miami FL,Loyola-Chicago
+2018,R1Y7,R2Y2,Weak,Y07,Y10,1305,1400,Nevada,Texas
+2018,R1Y8,R2Y1,Weak,Y08,Y09,1166,1243,Creighton,Kansas St
+2018,R1Z1,R2Z1,Strong,Z01,Z16,1462,1411,Xavier,TX Southern
+2018,R1Z2,R2Z2,Strong,Z02,Z15,1314,1252,North Carolina,Lipscomb
+2018,R1Z3,R2Z3,Strong,Z03,Z14,1276,1285,Michigan,Montana
+2018,R1Z4,R2Z4,Strong,Z04,Z13,1211,1422,Gonzaga,UNC Greensboro
+2018,R1Z5,R2Z4,Weak,Z05,Z12,1326,1355,Ohio St,S Dakota St
+2018,R1Z6,R2Z3,Weak,Z06,Z11,1222,1361,Houston,San Diego St
+2018,R1Z7,R2Z2,Weak,Z07,Z10,1401,1344,Texas A&M,Providence
+2018,R1Z8,R2Z1,Weak,Z08,Z09,1281,1199,Missouri,Florida St
+2018,R2W1,R3W1,Strong,R1W1,R1W8,,,,
+2018,R2W2,R3W2,Strong,R1W2,R1W7,,,,
+2018,R2W3,R3W2,Weak,R1W3,R1W6,,,,
+2018,R2W4,R3W1,Weak,R1W4,R1W5,,,,
+2018,R2X1,R3X1,Strong,R1X1,R1X8,,,,
+2018,R2X2,R3X2,Strong,R1X2,R1X7,,,,
+2018,R2X3,R3X2,Weak,R1X3,R1X6,,,,
+2018,R2X4,R3X1,Weak,R1X4,R1X5,,,,
+2018,R2Y1,R3Y1,Strong,R1Y1,R1Y8,,,,
+2018,R2Y2,R3Y2,Strong,R1Y2,R1Y7,,,,
+2018,R2Y3,R3Y2,Weak,R1Y3,R1Y6,,,,
+2018,R2Y4,R3Y1,Weak,R1Y4,R1Y5,,,,
+2018,R2Z1,R3Z1,Strong,R1Z1,R1Z8,,,,
+2018,R2Z2,R3Z2,Strong,R1Z2,R1Z7,,,,
+2018,R2Z3,R3Z2,Weak,R1Z3,R1Z6,,,,
+2018,R2Z4,R3Z1,Weak,R1Z4,R1Z5,,,,
+2018,R3W1,R4W1,Strong,R2W1,R2W4,,,,
+2018,R3W2,R4W1,Weak,R2W2,R2W3,,,,
+2018,R3X1,R4X1,Strong,R2X1,R2X4,,,,
+2018,R3X2,R4X1,Weak,R2X2,R2X3,,,,
+2018,R3Y1,R4Y1,Strong,R2Y1,R2Y4,,,,
+2018,R3Y2,R4Y1,Weak,R2Y2,R2Y3,,,,
+2018,R3Z1,R4Z1,Strong,R2Z1,R2Z4,,,,
+2018,R3Z2,R4Z1,Weak,R2Z2,R2Z3,,,,
+2018,R4W1,R5WX,Strong,R3W1,R3W2,,,,
+2018,R4X1,R5WX,Weak,R3X1,R3X2,,,,
+2018,R4Y1,R5YZ,Strong,R3Y1,R3Y2,,,,
+2018,R4Z1,R5YZ,Weak,R3Z1,R3Z2,,,,
+2018,R5WX,R6CH,Strong,R4W1,R4X1,,,,
+2018,R5YZ,R6CH,Weak,R4Y1,R4Z1,,,,
+2018,R6CH,R7WIN,Strong,R5WX,R5YZ,,,,
+2018,R7WIN,R6CH,,,,,,,

join_outputs_to_slots.py

@@ -0,0 +1,200 @@
+import pandas as pd
+import csv
+
+folder = 'data'
+results_folder = 'results'
+prediction_year = 2018
+# list to collect all print statements to go in a results file
+readable_outputs = []
+
+def format_pred_outputs( pred ):
+    '''
+    Take outputs from model.py and format them for joining to slot data
+    Return pred dataframe with added columns Year, TeamAID, TeamBID, PredIDConcat, ProbAOverB
+    '''
+    print("format prediction outputs")
+    formatted_data = []
+    headers = [ 'Year', 'TeamAID', 'TeamBID', 'PredIDConcat', 'ProbAOverB' ]
+    # for each row in pred data
+    for index, row in pred.iterrows():
+        # split id column, format is YYYY_TeamAID_TeamBID
+        id_split = row.id.split( '_' )
+
+        # get year
+        year = id_split[0]
+        # get team A (first one alphabetically)
+        teamA = id_split[1]
+        # get team B (second one alphabetically)
+        teamB = id_split[2]
+
+        # get win probability of team A beating team B
+        prob = row.pred
+
+        # save to new dataframe
+        row_data = [year, teamA, teamB, teamA + "_" + teamB, prob]
+        formatted_data.append(row_data)
+
+    # return dataframe
+    return pd.DataFrame(formatted_data, columns=headers)
+
+def get_readable_slot( slot_id ):
+    '''
+    Splits slot id into components Round, Region to get readable names
+    Returns readable string of the slot
+    '''
+    print("get readable slot name")
+    # rounds lookup
+    rounds = {
+        'R0': 'First Four',
+        'R1': 'First Round',
+        'R2': 'Second Round',
+        'R3': 'Sweet 16',
+        'R4': 'Elite Eight',
+        'R5': 'Final Four',
+        'R6': 'National Championship',
+    }
+    # region lookup
+    regions = {
+        'W': 'East',
+        'X': 'Midwest',
+        'Y': 'South',
+        'Z': 'West',
+        'WX': 'East/Midwest',
+        'YZ': 'South/West',
+        'CH': 'Final'
+    }
+
+    # get round from slot id split, first two letters
+    slot_round = slot_id[:2]
+    # get region from slot id split
+    slot_region = ''
+    if int(slot_id[1]) < 5:
+        # first 4 rounds, it's the third letter
+        slot_region = slot_id[2]
+    else:
+        # rounds 5 and 6, third and fourth letters
+        slot_region = slot_id[2:4]
+    # return combined region - round string
+    return regions[ slot_region ] + ' - ' + rounds[ slot_round ]
+
+def get_slot_winner( pred_df, slot_row ):
+    '''
+    For the given round, get the teams in the given slots
+    and then join to the team's probability of beating that other one
+    Return list of StrongTeamID, StrongTeamName
+    where StrongTeam = the team with the higher win probability
+    '''
+    print("get slot winner")
+    # create lookup of strong team ID by strong team name, also same for weak team
+    name_id_dict = {}
+    name_id_dict[slot_row.StrongTeamName] = slot_row.StrongTeamID
+    name_id_dict[slot_row.WeakTeamName] = slot_row.WeakTeamID
+
+    # concatenate them in the alphabetical order of StrongTeamName vs WeakTeamName
+    # to match the unique pred id, which is alpha sorted
+    alpha_teams = sorted([slot_row.StrongTeamName, slot_row.WeakTeamName])
+    teamA = alpha_teams[0]
+    teamB = alpha_teams[1]
+    SlotIDConcat = name_id_dict[ teamA ] + "_" + name_id_dict[ teamB ]
+
+    # get the ProbAOverB from pred_data
+    # where SlotIDConcat equals PredIDConcat
+    probA = pred_df[pred_df['PredIDConcat'] == SlotIDConcat]['ProbAOverB'].iloc[0]
+    readable_outputs.append(
+        [
+            'Chance that %s beats %s: %f' %
+            (teamA,teamB, probA)
+        ]
+    )
+    readable_outputs.append(
+        [
+            'Chance that %s beats %s: %f' %
+            (teamB,teamA, 1 - probA)
+        ]
+    )
+    if 0.39 <= probA <= 0.61:
+        readable_outputs.append(
+            ['***Close call!']
+        )
+
+    # save the highest probability for the next round
+    winner_data = []
+    # get readable slots from slot ID
+    slot_name = get_readable_slot( row.Slot )
+    if probA >= 0.5:
+        # if ProbAOverB is bigger, then Team A advances
+        readable_outputs.append(
+            [
+                '%s winner: %s (%f)' %
+                (slot_name, teamA, probA)
+            ]
+        )
+        winner_data = [name_id_dict[ teamA ], teamA]
+    else:
+        # else, Team B advances
+        readable_outputs.append(
+            [
+                '%s winner: %s (%f)' %
+                (slot_name, teamB, 1 - probA)
+            ]
+        )
+        winner_data = [name_id_dict[ teamB ], teamB]
+
+    # alert if weak team beats strong team
+    if slot_row.WeakTeamID == winner_data[0]:
+        readable_outputs.append(
+            ['*****Upset alert!']
+        )
+    return winner_data
+
+
+if __name__ == "__main__":
+    # slots data
+    slot_dtypes = {
+        'StrongTeamID': str,
+        'WeakTeamID': str
+    }
+    slots =  pd.read_csv(folder + '/NCAATourneySlots_Detailed_2018.csv', dtype=slot_dtypes)
+    slots.head()
+    # model prediction outputs
+    predictions =  pd.read_csv(results_folder + '/submission.csv')
+    predictions.head()
+
+    # format pred data to join with slots
+    pred_formatted = format_pred_outputs( predictions )
+    pred_formatted.head()
+
+    # for each slot
+    for row in slots.itertuples(index=False):
+        print("Joining slots to predictions")
+        readable_outputs.append(
+            ['------------------------------------------']
+        )
+        if row.Slot == 'R7WIN':
+            readable_outputs.append(
+                [
+                    'Overall 2018 champion: %s' %
+                    (row.StrongTeamName)
+                ]
+            )
+            break
+        # join slots and pred to get higher prob team
+        # returns [StrongTeamID, StrongTeamName]
+        slot_winner = get_slot_winner(pred_formatted, row)
+
+        # save higher team in appropriate slot for next round
+        next_slot = row.NextSlot
+        seed_type = row.NextSeed
+
+        # assign the updated winner data in the next slot
+        slots.loc[slots['Slot'] == next_slot, [seed_type + 'TeamID',seed_type + 'TeamName']] = slot_winner[0], slot_winner[1]
+
+    slots.tail()
+    # output updated slot data to csv in results folder
+    print("Write updated slots data to file")
+    slots.to_csv(results_folder + '/less-readable-predictions-by-slots.csv', index=False)
+    # create readable results csv
+    print("Writing %d readable bracket results." % len(slots))
+    with open(results_folder + '/readable-predictions-by-slots.csv', 'w') as f:
+        writer = csv.writer(f)
+        writer.writerows(readable_outputs)