def main():
with open('smartphone_classifier', 'rb') as classifier:
# Load the trained model and X and y values
model = pickle.load(classifier)
X, y = load_dataset()
# Classify as 'smartphone' or 'nao-smartphone'
y_pred = model.predict(X)
# Print the results
print("Matriz de confusao: ")
print(confusion_matrix(y, y_pred))
print("Resumo:")
print(classification_report(y, y_pred))
print("Acuracia:", round((accuracy_score(y, y_pred) * 100), 4), "%")
generate_output_file(y_pred)
return True
Format of the input configuration: additional features. Possible options:
-BASE: no changes; simple count
-TD: adds time dependency in the count
-EMI: mutual information extened matrix (sensor event window adjency)
-TD+EMI: combination of TD and EMI
-ALL: iteration that serves to test all possible input configurations
"""
if __name__ == "__main__":
#Reading the inputs
if len(sys.argv) < 4:
#Bad usage of the command
msg_error(USAGE)
#Selecting the dataset
dataset, dataset_name = load_dataset(sys.argv[3])
#Divide the testing and training data
try:
int_data = dataset.obtaining_data()
except Exception as exc:
msg_error(exc)
#Selecting which features to test
if sys.argv[1] == "ALL":
features = POSIBLE_FEATURE_CONFIG.copy()
elif sys.argv[1] in POSIBLE_FEATURE_CONFIG:
features = [sys.argv[1]]
else:
#Unkown feature
msg_error("Unkown feature: " + sys.argv[1])