def make_prediction():
new_data = scrape()
DATA.append(new_data)
X, feature_names = feature_engineering(DATA, [])
proba = MODEL.predict_proba(X[-1].reshape(1,X.shape[1]))[:,1][0]
probability = "{:.2%}".format(proba)
if proba < 0.2:
fraud_text = "Pas de Fraude!"
else:
fraud_text = "Fraude!"
return render_template('index.html', prediction=fraud_text,
proba=probability, event_data=new_data)
def make_predictions():
new_data = scrape()
DATA.append(new_data)
X, feature_names = feature_engineering(DATA, [])
proba = MODEL.predict_proba(X[-1].reshape(1,X.shape[1]))[:,1][0]
probability = "{:.2%}".format(proba)
if proba < 0.2:
fraud_text = "Pas de Fraude!"
else:
fraud_text = "Fraude !"
document = {"_id":new_data['object_id'],
"response":new_data,
"prediction":fraud_text,
"probability":proba}
try:
coll.insert_one(document)
except:
pass
return fraud_text, probability, new_data
import numpy as np
import pandas as pd
import pickle
from sklearn.metrics import recall_score, precision_score, f1_score, confusion_matrix
def load_classifier(filename):
with open(filename, "rb") as clf_infile:
clf = pickle.load(clf_infile)
return clf
if __name__ == '__main__':
## get training features
with open('data/X_test.json') as f:
data = json.load(f)
X, feature_names = feature_engineering(data)
## get training labels
with open('data/y_test.json') as f:
labels = json.load(f)
df_y = pd.DataFrame(labels)
# get y
y = df_y.pop('fraud').values
clf = load_classifier("our_classifier.pkl")
predictions = clf.predict(X)
print(confusion_matrix(y, predictions))
print("Recall: {:.3%}".format(precision_score(y, predictions)))
print("Precision: {:.3%}".format(recall_score(y, predictions)))
print("F1: {:.3%}".format(f1_score(y, predictions)))
