Introduction
For many large cities, air pollution is a severe problem. We examine machine learning algorithms that can accurately predict air quality concentrations in a city. Air quality data are essential to determine which areas contain high concentrations of pollutants. Accurate predictions of air pollutants can help governments and the general public make informed decisions.
![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F1249985%2Fe7b1077c038fb27520fc6804cec7b181%2Faq.png?generation=1603461646664880&alt=media)
Data
We will be examining multiple kinds of pollutants: SO2, NO2, O3, CO, and PM2.5 (particulate matter with a diameter of 2.5 micrometers or less). Specifically, we will be predicting PM2.5 concentrations using the concentrations of the other pollutants. All units are in µg/m3 (microgram/cubic meter).
Problem statement
The goal is to predict the PM2.5 concentration from SO2, NO2, O3, and CO concentrations. You will implement various baseline methods, and then you can extend your implementations in order to improve the performance of the machine learning algorithms for extra credit.  The evaluation metric is Mean Squared Error (MSE) between the predicted PM2.5 concentrations and the ground-truth PM2.5 concentrations.
Acknowledgements
The air quality data are from the Beijing Municipal Environmental Monitoring Center. We thank the EECS 545 instructors for additional feedback for this assignment.