def train_ada_boost(X, Y, estimators, classes, features):
train_x = X[0:3600, :]
train_y = Y[0:3600]
validate_x = X[3600:, :]
validate_y = Y[3600:]
clf = AdaBoostClassifier(n_estimators=estimators)
clf.n_classes_ = classes
clf.n_features_ = features
clf.fit(train_x, train_y)
predictions = clf.predict(validate_x).reshape(400, 1)
result = sum([1 if predictions[i] == validate_y[i] else 0 for i in range(validate_y.shape[0])]) / \
validate_y.shape[0] * 100
return result
def main():
# Prepare data
df_even = pd.read_csv('DNN/adversary_even_log.csv', index_col=0)
df_even = df_even.loc[df_even['Class']==0]
df_even = df_even.loc[df_even['generator']!=2]
df_even = df_even.loc[df_even['nTags']==2]
df_odd = pd.read_csv('DNN/adversary_odd_log.csv', index_col=0)
df_odd = df_odd.loc[df_odd['Class']==0]
df_odd = df_odd.loc[df_odd['generator']!=2]
df_odd = df_odd.loc[df_odd['nTags']==2]
df_odd[variables] = scale(df_odd[variables])
df_even[variables] = scale(df_even[variables])
# construction of bdt
bdt_even = AdaBoostClassifier(DecisionTreeClassifier(max_depth=max_depth, min_samples_leaf=0.01),
learning_rate=learning_rate,
algorithm="SAMME",
n_estimators=n_estimators
)
bdt_even.n_classes_ = 2
bdt_odd = AdaBoostClassifier(DecisionTreeClassifier(max_depth=max_depth, min_samples_leaf=0.01),
learning_rate=0.15,
algorithm="SAMME",
n_estimators=n_estimators
)
bdt_odd.n_classes_ = 2
# Convert generator class to categorical
z_even = to_categorical(df_even['generator'], num_classes=2)
z_odd = to_categorical(df_odd['generator'], num_classes=2)
# fitting to generators
bdt_even.fit(df_even[variables], df_even['generator'], sample_weight=df_even['EventWeight'])
bdt_odd.fit(df_odd[variables], df_odd['generator'], sample_weight=df_odd['EventWeight'])
# Scoring
df_odd['bdt_outcome'] = bdt_odd.decision_function(df_odd[variables]).tolist()
df_even['bdt_outcome'] = bdt_even.decision_function(df_even[variables]).tolist()
print(bdt_odd.score())
df = pd.concat([df_odd,df_even])
# plotting BDT outcome for different generators
gen1 = df.loc[df['generator']==0]
gen2 = df.loc[df['generator']==1]
# plt.hist(gen1['bdt_outcome'],bins=70,color='red',alpha=0.5,density=True)
# plt.hist(gen2['bdt_outcome'],bins=70,color='blue',alpha=0.5,density=True)
#
# plt.show()
# calculating fpr, tpr and auc for roc curve
fpr = dict()
tpr = dict()
area = dict()
z_all = to_categorical(df['generator'], num_classes=2)
fpr[0], tpr[0], _ = roc(z_all[:,0], df['bdt_outcome'],sample_weight=df['EventWeight'])
# area[0] = auc(fpr[0], tpr[0])
# plotting the roc curve
# plt.plot(fpr[0], tpr[0], color='darkorange',lw=1,label='PYTHIA, ROC curve (area = %0.2f)' % area[0])
# plt.plot([0, 1], [0, 1], color='navy', lw=1, linestyle='--')
# plt.legend()
# plt.savefig('roc.png', bbox_inches='tight')
# plt.show(block=True)
print('It runs. ')
