def get(self): subject = 'Subscribers.csv from ' + self.request.host mail.send_mail(sender=config.ADMIN, to=config.ADMIN, subject=subject, body=util.get_csv())
def get(self): self.response.headers['Content-Type'] = 'text/plain' self.response.out.write(util.get_csv())
def run(gamesToPredict):
raw_data = util.get_csv()
# gamesToPredict = 4
# target data
home_target = raw_data["Home_Score"]
away_target = raw_data["Away_Score"]
# data to analyse
home_data = [
'WDW_ Home', 'PRDZ_Home', 'FORE_Home', 'VITI_Home',
'FREESUPERTIPS_Home', 'SCOREPREDICTOR_Home'
]
away_data = [
"WDW_Away", "PRDZ_Away", "FORE_Away", "VITI_Away",
"FREESUPERTIPS_Away", "SCOREPREDICTOR_Away"
]
home_var = raw_data[home_data]
away_var = raw_data[away_data]
# only played games to build the model
gamesPlayedIndex = home_target.count()
home_target = home_target.iloc[:gamesPlayedIndex, ]
away_target = away_target.iloc[:gamesPlayedIndex, ]
home_pred = home_var.iloc[gamesPlayedIndex:gamesPlayedIndex +
gamesToPredict, ]
away_pred = away_var.iloc[gamesPlayedIndex:gamesPlayedIndex +
gamesToPredict, ]
home_var = home_var.iloc[:gamesPlayedIndex, ]
away_var = away_var.iloc[:gamesPlayedIndex, ]
currentGame = raw_data[gamesPlayedIndex:gamesPlayedIndex + gamesToPredict]
for i in range(0, gamesToPredict):
print "\n " + currentGame['Home'].values[i] + " vs " + currentGame[
'Away'].values[i]
# models
# random forest
rf_home_scoreModel = RandomForestRegressor(n_estimators=100,
criterion='mae').fit(
home_var, home_target)
rf_away_scoreModel = RandomForestRegressor(n_estimators=100,
criterion='mae').fit(
away_var, away_target)
rf_home_pred = rf_home_scoreModel.predict(home_pred)
rf_away_pred = rf_away_scoreModel.predict(away_pred)
print "\nRandom-Forest: \n\t", int(round(rf_home_pred[i])), "-", int(
round(rf_away_pred[i])), "\n w/o round: ", round(
rf_home_pred[i], 2), "-", round(rf_away_pred[i], 2)
# LinearRegression
lr_home_scoreModel = LinearRegression().fit(home_var, home_target)
lr_away_scoreModel = LinearRegression().fit(away_var, away_target)
lr_home_pred = lr_home_scoreModel.predict(home_pred)
lr_away_pred = lr_away_scoreModel.predict(away_pred)
print "\nLinear-Regression:\n\t", int(round(
lr_home_pred[i])), "-", int(
round(lr_away_pred[i])), "\n w/o round: ", round(
lr_home_pred[i], 2), "-", round(lr_away_pred[i], 2)
# nn
# def baseline_model():
# nn = Sequential()
# nn.add(Dense(10, input_dim=6, activation='relu'))
# nn.add(Dense(1))
# nn.compile(loss='mean_squared_error', optimizer='adam')
# return nn
# nn_est = KerasRegressor(build_fn=baseline_model, epochs = 100, batch_size=3)
# nn_est.fit(home_var, home_target)
# nn_est.predict(home_pred)
# print(nn_est.predict(home_pred))
print("\nnn-model:")
model = Sequential()
model.add(Dense(4, input_dim=6, activation='relu'))
model.add(Dense(1, activation='relu'))
model.add(Dense(1, activation='linear'))
model.compile(loss='mse', optimizer='adam')
model.fit(home_var, home_target, epochs=300, verbose=0)
nn_home = model.predict(home_pred)
model.fit(away_var, away_target, epochs=300, verbose=0)
nn_away = model.predict(away_pred)
f_home = int(round(float(str(nn_home[i][0]))))
f_away = int(round(float(str(nn_away[i][0]))))
print "\t" + str(f_home) + " - " + str(f_away), "\n w/o round: ", str(
round(nn_home[i][0], 2)), "-", str(round(nn_away[i][0], 2))
winnerPredict.prepare_data(raw_data)
print "\n----------------"