model = MLPClassifier(alpha=c, max_iter=100000) b = pd.DataFrame() b["binSpread"] = y_cvTest model.fit(X=x_cvTrain, y=y_cvTrain) for p in model.predict_proba(x_cvTest): if (model.classes_[1] == 1): predictions.append(p[1]) else: predictions.append(p[0]) b["PFITS"] = predictions predictions = [] allSums.append( testClassification(b, 300, 'Kelly', Year=False, Week=False, odds=-105, betType="Spread", printOption=False)) allScores.append(model.score(x_cvTest, y_cvTest)) counter += 1 print(c, np.average(allScores), np.average(allSums)) if (np.average(allScores) > best): bestC = c best = np.average(allScores) allScores = [] allSums = [] counter = 0 print("The best C was", bestC) model = MLPClassifier(max_iter=100000, alpha=bestC) b = pd.DataFrame()
import pandas as pd import numpy as np from cfbFcns import standardizeTeamName from evalPredictions import testClassification #Spread Part a = pd.read_csv( './csv_Data/LogisticPythonPredictionsLRWeek5+Leave-One-OutNoStepwiseScaled.csv', encoding="ISO-8859-1") testClassification(a, 300, 'Kelly', Year=True, Week=True, odds=-105, betType="Spread") a = pd.read_csv( './csv_Data/LogisticPythonPredictionsLRWeek5+Leave-One-OutNoStepwiseScaledC0.25.csv', encoding="ISO-8859-1") testClassification(a, 300, 'Kelly', Year=True, Week=True, odds=-105, betType="Spread") a = pd.read_csv( './csv_Data/LogisticPythonPredictionsLRWeek5+Leave-One-OutNoStepwiseScaledC0.1.csv', encoding="ISO-8859-1")
model = LogisticRegression(max_iter=100000) X_train, X_test, y_train, y_test = train_test_split(temp, Y, test_size=0.33, random_state=42) model.fit(X=X_train, y=y_train) for p in model.predict_proba(X_test): if (model.classes_[1] == 1): predictions.append(p[1]) else: predictions.append(p[0]) b = pd.DataFrame() b["binTotal"] = y_test b["PFITS"] = predictions predictions = [] netWin = testClassification(b, 300, 'Standard', betType="O/U") best = netWin bestCols = [] removedCols = [] noImprovement = True colCount = 0 removedCount = 0 bestImproved = 0 for col in temp.columns: bestCols.append(col) colCount += 1 random.shuffle(bestCols) X = temp.copy() print("Starting Removal of Features") while (bestImproved < colCount): print(bestImproved, colCount)
# for p in model.predict_proba(x1.reshape(-1,1)): # if (model.classes_[1] == 1): # predictions.append(p[1]) # if (p[1] > 0.5122): # z.append(p[1]) # else: # predictions.append(p[0]) bSpread["PFITS"] = x bSpread["binSpread"] = y print( "SPREAD ------------------------------------------------------------------------" ) testClassification(bSpread, 300, 'Kelly', Year=False, Week=False, odds=-105, betType="Spread", printOption=True) print( "-------------------------------------------------------------------------------" ) predictions = [] allSums = [] x = [] y = [] z = [] allScores = [] a = pd.read_csv('./new_csv_Data/bigboyBinClassificationTotalsALTTrain.csv', encoding="ISO-8859-1")
trainRows.append(j) tempTrainX = X_train.iloc[trainRows] tempTrainY = y_train.iloc[trainRows] tempValX = X_train.iloc[valRows] tempValY = y_train.iloc[valRows] b = pd.DataFrame() b["binSpread"] = tempValY model.fit(X = tempTrainX, y = tempTrainY) for p in model.predict_proba(tempValX): if (model.classes_[1] == 1): predictions.append(p[1]) else: predictions.append(p[0]) b["PFITS"] = predictions predictions = [] cvSum += testClassification(b, 300, 'Kelly', Year = False, Week = False, odds = -105, betType = "Spread", print = False) allSums.append(cvSum) cvSum = 0 print ("Initial cvSum =", np.average(allSums)) best = np.average(allSums) bestCols = [] removedCols = [] noImprovement = True colCount = 1 removedCount = 0 bestImproved = 0 for col in xCols: if ("Elo" not in col): removedCols.append(col) removedCount += 1 bestCols.append("Elo_diff_aboveAvg")
best = 0 bestCols = [] temp = X.copy() model = LogisticRegression(max_iter = 100000) X_train, X_test, y_train, y_test = train_test_split(temp, Y, test_size=0.33, random_state=42) model.fit(X = X_train, y = y_train) for p in model.predict_proba(X_test): if (model.classes_[1] == 1): predictions.append(p[1]) else: predictions.append(p[0]) b = pd.DataFrame() b["binSpread"] = y_test b["PFITS"] = predictions predictions = [] netWin = testClassification(b, 300, 'Standard') best = netWin bestCols = [] removedCols = [] noImprovement = True colCount = 0 removedCount = 0 bestImproved = 0 for col in temp.columns: bestCols.append(col) colCount += 1 random.shuffle(bestCols) X = temp.copy() print ("Starting Removal of Features") while (bestImproved < colCount): print (bestImproved, colCount)