def Eval(self, test):
Y_pred = self.PredictLabel(test.X)
ev = Eval(Y_pred, test.Y)
print("For Positive Class:")
print("Test Accuracy: ",ev.Accuracy())
print("Test Recall: ",ev.Recall())
print("Test Precision: ",ev.Precision())
print("\n")
print("For Negative Class:")
ev_neg = Eval([1 if i == -1 else -1 for i in Y_pred], [1 if i == -1 else -1 for i in test.Y])
print("Test Accuracy: ",ev_neg.Accuracy())
print("Test Recall: ",ev_neg.Recall())
print("Test Precision: ",ev_neg.Precision())
probality_threshold=[0.2,0.4,0.6,0.8]
Precision=[]
Recall=[]
Precision.append(ev.Precision())
Recall.append(ev.Recall())
length=len(probality_threshold)
for i in range(0,length):
Y_pred = self.PredictLabel(test.X,probality_threshold[i])
ev = Eval(Y_pred, test.Y)
Precision.append(ev.Precision())
Recall.append(ev.Recall())
plt.plot(Precision,Recall)
plt.ylabel('Recall')
plt.xlabel('Precision')
#plt.ylim([0.0, 1.05])
#plt.xlim([0.0, 1.0])
plt.title('2-class Precision-Recall curve')
plt.show()
def Eval(self, test):
Y_pred = self.PredictLabel(test.X)
ev = Eval(Y_pred, test.Y)
print("For Positive Class:")
print("Test Accuracy: ", ev.Accuracy())
print("Test Recall: ", ev.Recall())
print("Test Precision: ", ev.Precision())
print("\n")
print("For Negative Class:")
ev_neg = Eval([1 if i == -1 else -1 for i in Y_pred],
[1 if i == -1 else -1 for i in test.Y])
print("Test Accuracy: ", ev_neg.Accuracy())
print("Test Recall: ", ev_neg.Recall())
print("Test Precision: ", ev_neg.Precision())
ev.PvRcurve()
def Eval(self, test):
Y_pred = self.PredictLabel(test.X,0,True)
recall_pos = []
recall_neg = []
precision_pos = []
precision_neg =[]
ev = Eval(Y_pred, test.Y)
accy = ev.Accuracy()
#print(ev.EvalRecall())
#print(ev.EvalPrecision())
probThresh = [0.2,0.4,0.6,0.8]
for i in range(len(probThresh)):
pred = self.PredictLabel(test.X,probThresh[i],False)
ev = Eval(pred, test.Y)
#ev.EvalRecall()
#print('Threshold Value %f'%probThresh[i])
recallP,recallN=ev.EvalRecall()
recall_pos.append(recallP)
recall_neg.append(recallN)
precisionP,precisionN=ev.EvalPrecision()
precision_pos.append(precisionP)
precision_neg.append(precisionN)
plt.pause(0.1)
plt.xlabel('Recall')
plt.ylabel('Precision')
fig=plt.figure()
plt.title('For +1 Label')
plt.plot(recall_pos,precision_pos,'r')
fig.savefig('Pos_class.png')
plt.title('For -1 Label')
fig1=plt.figure(1)
plt.plot(recall_neg,precision_neg,'b')
fig1.savefig('Neg_class.png')
return accy
def plotLimitGraph(self, test):
x_axis = []
accuracy = []
for i in range(9):
Y_pred = self.PredictLabel(test.X, (i + 1) / 10)
ev = Eval(Y_pred, test.Y)
accuracy.append(ev.Accuracy())
x_axis.append((i + 1) / 10)
# Y_pred = self.PredictLabel(test.X, (i+1)/10)
# ev = Eval(Y_pred, test.Y)
# accuracy.append(ev.Accuracy())
# #Y_pred1 = np.array(Y_pred)
# #recall_pos.append(recall_score(test.Y,Y_pred1))
# #precision_pos.append( precision_score(test.Y,Y_pred1))
# #Y_pred_neg = np.array([1 if i == -1 else -1 for i in Y_pred])
# #Y_test_neg = np.array([1 if i == -1 else -1 for i in test.Y])
# #recall_neg.append(recall_score(Y_test_neg,Y_pred_neg))
# #precision_neg.append(precision_score(Y_test_neg,Y_pred_neg))
# x_axis.append((i+1)/10)
# # print(i,recall_pos,precision_pos)
plt.title('Recall Positive Graph.')
plt.plot(x_axis, self.pos_recall, label="Recall Positive")
plt.plot(x_axis, self.neg_recall, label="Recall Negative")
plt.xlabel('Threshold')
plt.ylabel('Recall')
plt.title('Recall Plot')
plt.legend()
plt.show()
plt.plot(x_axis, self.pos_precision, label="Precision Positive")
plt.plot(x_axis, self.neg_precision, label="Precision Negative")
plt.xlabel('Threshold')
plt.ylabel('Precision')
plt.title('Precision Plot')
plt.legend()
plt.show()
def Eval(self, X_test, Y_test):
Y_pred = self.Predict(X_test)
ev = Eval(Y_pred, Y_test)
return ev.Accuracy()
X_train=X_train.drop('attacking_work_rate',1)
X_test=X_test.drop('preferred_foot',1)
X_train=X_train.drop('preferred_foot',1)
from sklearn.naive_bayes import GaussianNB
gnb = GaussianNB()
model= gnb.fit(X_train, y_train)
y_pred = gnb.fit(X_train, y_train).predict(X_test)
y_pred.dtype
model.score(X_test,y_test)
print("Number of mislabeled points out of a total %d points : %d" % (X_test.shape[0],(y_test != y_pred).sum()))
from Eval import Eval
eval1 = Eval(y_pred, np.array(y_test))
print("Positive Class:")
print("Accuracy: ",eval1.Accuracy())
from sklearn.metrics import recall_score,precision_score,accuracy_score
print(recall_score(y_test,y_pred,average=None))
print(precision_score(y_test,y_pred,average=None))
print(accuracy_score(y_test,y_pred))
import pickle
filename = 'finalized_model.sav'
pickle.dump(model, open(filename, 'wb'))
loaded_model = pickle.load(open(filename, 'rb'))
loaded_model.score(X_test,y_test)
#X_test.to_csv('sample_players1.csv')
#y_test.to_csv('pred.csv')
#X_test.iloc[:,3:39].head(1)
#loaded_model.theta_.shape
#X_test.shape
def EvalProbabilities(self, test, indexes):
Y_pred = self.PredictProb(test, indexes)
ev = Eval(Y_pred, test.Y[indexes])
return ev.Accuracy()
def EvalLabels(self, test):
Y_pred = self.PredictLabel(test.X)
ev = Eval(Y_pred, test.Y)
return ev.Accuracy()
def Eval(self, test):
#With the labels obtained and those that really are we obtain the accuracy
Y_pred = self.PredictLabel(test.X)
ev = Eval(Y_pred, test.Y)
return ev.Accuracy()
def Eval(self, test):
y_pred = self.predictlabel(test.X)
ev = Eval(y_pred, test.Y)
return ev.Accuracy()
def Eval(self, X_test, Y_test):
print "testing..."
Y_pred = self.Predict(X_test)
ev = Eval(Y_pred, Y_test)
return ev.Accuracy()
def Eval(self, test):
print("Predicting test reviews labels ...")
Y_pred = self.PredictLabel(test.X)
print("Evaluate the NaiveBayes model ...")
ev = Eval(Y_pred, test.Y)
return ev.Accuracy()
def Eval(self, test):
Y_pred = self.PredictLabel(test.X)
# Y_pred = self.PredictLabel(test,len(test))
ev = Eval(Y_pred, test.Y)
# ev = Eval(Y_pred.pred, test.Y)
return [ev.Accuracy(), ev.EvalPrecition(), ev.EvalRecall()]
nb = NaiveBayes(traindata, float(sys.argv[2]))
for i in range(10):
res = nb.PredictProb(testdata, i)
print(i, res)
print("Predict Label with Threshold")
precition = []
recall = []
for i in np.arange(0.1, 1.0, 0.1):
pred = nb.PredictLabelTh(testdata, i)
ev = Eval(pred, testdata.Y)
precition.append(ev.EvalPrecition())
recall.append(ev.EvalRecall())
# ev = Eval(Y_pred.pred, test.Y)
print('Threshold:', i, ' Accuracy: ', ev.Accuracy(), ' Precition: ',
ev.EvalPrecition(), ' Recall: ', ev.EvalRecall())
'''
# Create some mock data
t = recall
data1 = precition
data2 = np.arange(0.05, 1.0, 0.05)
fig, ax1 = plt.subplots()
color = 'tab:blue'
ax1.set_xlabel('Recall')
ax1.set_ylabel('Precition', color=color)
ax1.plot(t, data1, color=color)
ax1.tick_params(axis='y', labelcolor=color)
ax2 = ax1.twinx()
color = 'tab:red'
ax2.set_ylabel('Threshold', color=color)
def Eval(self, test):
print "testing..."
Y_pred = self.PredictLabel(test.X)
ev = Eval(Y_pred, test.Y)
return ev.Accuracy()