predict_fine.py [download]

#!/usr/bin/env python3

import sklearn
import joblib
import sys
import os

if len(sys.argv) != 2:
    print("usage: {} csv_file".format(sys.argv[0]))
    sys.exit(1)

test_filename = sys.argv[1]
if not os.path.exists(test_filename):
    print("usage: {} csv_file".format(sys.argv[0]))
    print("{} does not exist.".format(test_filename))
    sys.exit(1)
    

# read data, define fields, etc.
from showcase_common_fine import *
test_data = pd.read_csv(test_filename, index_col=0)
X_test = test_data[feature_names]
y_test = test_data[label_name]

# load model
(regressor,scaler) = joblib.load(model_filename)

# scale data with already fit scaler
X_test = scaler.transform(X_test)

# ask model to score the data
score_test = regressor.score(X_test, y_test)
print("R^2: {}".format(score_test))

# show mean square error and mean absolute error
y_predicted = regressor.predict(X_test)
loss_test = sklearn.metrics.mean_squared_error(y_test, y_predicted)
print("MSE: {}".format(loss_test))
loss_test = sklearn.metrics.mean_absolute_error(y_test, y_predicted)
print("MAE: {}".format(loss_test))

print()
print()
for j in range(len(y_predicted)):
    print(X_test[j], y_test.iloc[j], y_predicted[j])

Last Updated 01/23/2025