import pandas as pd

import dataframes as df

import analytics as a

# Useful data

# Wins

# Championships

# Podiums

# average Positions gained

# Points

# Fastest laps

# Pole positions

drivers = df.drivers_df()

drivers = drivers[['driverId','Full Name']]

results = df.results_df

wins = results[results['positionOrder'] == 1]

podiums = results[results['positionOrder'] <= 3]

wins = wins.groupby('driverId').size().reset_index(name='wins')

podiums = podiums.groupby('driverId').size().reset_index(name='podiums')

drivers = drivers.merge(podiums, on='driverId', how='left')

drivers = drivers.merge(wins, on='driverId', how='left')

drivers['wins'] = drivers['wins'].fillna(0)

drivers['podiums'] = drivers['podiums'].fillna(0)

championships = df.driver_standings_df

championships = championships.join(df.races_df(), on='raceId', lsuffix='l', rsuffix='r')

championships = championships[championships['position'] == 1]

championships = championships.loc[championships.groupby('year')['round'].idxmax()]

drivers['championships'] = championships.groupby('driverId').size().reset_index(name='championships')['championships']

drivers['championships'] = drivers['championships'].fillna(0)

# points = results.groupby('driverId')['points'].sum().reset_index(name='points')

# drivers = drivers.merge(points, on='driverId', how='left')

qualifying_results = df.qualifying_results_df

pole_positions = qualifying_results[qualifying_results['position'] == 1]

pole_positions = pole_positions.groupby('driverId').size().reset_index(name='pole_positions')

drivers = drivers.merge(pole_positions, on='driverId', how='left')

drivers['pole_positions'] = drivers['pole_positions'].fillna(0)

finish_positions_vs_start = results[['driverId', 'grid', 'positionOrder']]

finish_positions_vs_start['positionDiff'] = finish_positions_vs_start['grid'] - finish_positions_vs_start['positionOrder']

average_position_diff = finish_positions_vs_start.groupby('driverId')['positionDiff'].mean().reset_index(name='average_position_diff')

drivers = drivers.merge(average_position_diff, on='driverId', how='left')

fastest_laps = results[results['fastestLapTime'].notnull()]

fastest_laps = fastest_laps.groupby('raceId')['fastestLapTime'].min()

fastest_laps = fastest_laps.reset_index().merge(results[['raceId', 'driverId', 'fastestLapTime']], on=['raceId', 'fastestLapTime'])

drivers = drivers.merge(fastest_laps.groupby('driverId').size().reset_index(name='fastest_laps'), on='driverId', how='left')

drivers['fastest_laps'] = drivers['fastest_laps'].fillna(0)

print(drivers)

def create_driver_pairs(race_results, drivers):

pairs = []

for race_id in race_results['raceId'].unique():

race_drivers = race_results[race_results['raceId'] == race_id]['driverId'].values

for i in range(len(race_drivers)):

for j in range(len(race_drivers)):

driver1 = race_drivers[i]

driver2 = race_drivers[j]

driver1_features = drivers[drivers['driverId'] == driver1].drop(columns=['driverId','Full Name']).values.flatten()

driver2_features = drivers[drivers['driverId'] == driver2].drop(columns=['driverId','Full Name']).values.flatten()

feature_diff = driver1_features - driver2_features

res = race_results[race_results['raceId'] == race_id]

# print(int(res[res['driverId'] == driver1]['positionOrder'].values[0]))

# print(int(res[res['driverId'] == driver2]['positionOrder'].values[0]))

winner = 1 if int(res[res['driverId'] == driver1]['positionOrder'].values[0]) <= int(res[res['driverId'] == driver2]['positionOrder'].values[0]) else 2

# print(winner)

pairs.append(list(feature_diff) + [winner])

return pd.DataFrame(pairs, columns=['podiums','wins','championships','pole_positions','average_position_diff','fastest_laps'] + ['winner'])

time_start = pd.Timestamp.now()

race_results = df.results_df[['raceId','driverId','positionOrder']]

print(race_results)

pairs = create_driver_pairs(race_results, drivers)

time_fin = pd.Timestamp.now()

print("Time taken:", (time_fin - time_start))

X = pairs.drop(columns=['winner'])

Y = pairs['winner']

print(pairs[pairs['winner'] == 1].shape)

print(pairs[pairs['winner'] == 2].shape)

print(pairs.shape)

X_train = X[0::2]

Y_train = Y[0::2]

X_test = X[1::2]

Y_test = Y[1::2]

from sklearn.linear_model import LogisticRegression

model = LogisticRegression()

model.fit(X_train, Y_train)

y_pred = model.predict(X_test)

from sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score

accuracy = accuracy_score(Y_test, y_pred)

precision = precision_score(Y_test, y_pred)

recall = recall_score(Y_test, y_pred)

f1 = f1_score(Y_test, y_pred)

print(f'Accuracy: {accuracy}')

print(f'Precision: {precision}')

print(f'Recall: {recall}')

print(f'F1 Score: {f1}')

def predict_winner(driver1_id, driver2_id, drivers, model):

driver1_features = drivers[drivers['driverId'] == driver1_id].drop(columns=['driverId', 'Full Name']).values.flatten()

driver2_features = drivers[drivers['driverId'] == driver2_id].drop(columns=['driverId', 'Full Name']).values.flatten()

feature_diff = driver1_features - driver2_features

print(feature_diff)

prediction = model.predict([feature_diff])

return driver1_id if prediction == 1 else driver2_id

import joblib

joblib.dump(model, 'trained_model.pkl')