The data set is from kaggle. It records 18 air features in 24 hours for a entire year. The goal is to predict the PM2.5 for the eleventh hour with previous 10 hours fearture data.
I already edited the feature and hours, and only used 12 features and 8 hours out of 10. I applied both the linear regression model and non-linear model. For the updating of gradient descent, I used the adagrad method.
To run the code, download the py file and both x.npy and y.npy.
python3 linear_ regression.py