return self.model.predict(test1).tolist()
def save(self, f_name):
dump(self.normalizer, f_name + '_normalizer.joblib')
dump(self.model, f_name + '_model.joblib')
def load_model(self, f_name):
self.normalizer = load(f_name + '_normalizer.joblib')
self.model = load(f_name + '_model.joblib')
if __name__ == '__main__':
#read train data from train.csv
train_df = pd.read_csv('../data/train.csv',
usecols=[
'STAT_CAUSE_DESCR', 'LATITUDE', 'LONGITUDE',
'DISCOVERY_DATE', 'FIRE_SIZE'
])
y = pd.DataFrame()
y['STAT_CAUSE_DESCR'] = train_df['STAT_CAUSE_DESCR']
#create labels by grouping the causes
y = labels.createLabel(y)
y = y['STAT_CAUSE_DESCR'].astype(int)
train_df = train_df.drop(columns=['STAT_CAUSE_DESCR'])
#train and save the model
model = LRModel()
model.trainModel(train_df, y)
model.save("../models/lrmodel")
from LogisticRegression import LRModel
from collections import Counter
from sklearn.metrics import f1_score,confusion_matrix,accuracy_score
if __name__ == '__main__':
test_df = pd.read_csv('../data/test.csv',usecols = ['STAT_CAUSE_DESCR','LATITUDE','LONGITUDE','DISCOVERY_DATE','FIRE_SIZE'])
#Separate the labels
y_test = pd.DataFrame()
y_test['STAT_CAUSE_DESCR']=test_df['STAT_CAUSE_DESCR']
test_df = test_df.drop(columns=['STAT_CAUSE_DESCR'])
#create new classes
y_test=labels.createLabel(y_test)
y_test=y_test['STAT_CAUSE_DESCR'].astype(int)
#Predict using Logistic Regression trained model
model = LRModel()
model.load_model("../models/lrmodel")
y_pred = model.predict(test_df)
#Evaluate#
print("Logistic Regression Model")
print("------------------------------------------------------------------------------------------")
print("F1-score: %f"%(f1_score(y_test,y_pred,average='macro')))
print("Accuracy Score: %f"%(accuracy_score(y_test,y_pred)))
print(confusion_matrix(y_test, y_pred))