Share on Facebook Tweet on Twitter Share on LinkedIn Share by email
Urban Air
Urban Air

Using a diversity of big data to infer and predict fine-grained air quality throughout a city, and finally tackle air pollutions.


Many countries are suffering from air pollutions. Cities have built a few air quality monitoring stations to inform people urban air quality every hour. Influenced by multiple complex factors, however, urban air quality is highly skewed in a city, varying by locations significantly and changing over time differently in different places. Thus, we cannot infer the air quality of a location without a monitoring station by simply using a linear interpolation.

Step 1:

The first step of this project is to infer the real-time and fine-grained air quality of arbitrary location by using two parts of data. One is the real-time and historical air quality data from existing monitoring stations. The other is five additional data sources we observed in a city, consisting of meteorological data, traffic, human mobility, POIs, and road network data. We propose a semi-supervised learning approach based on a co-training framework that consists of two separated classifiers. One is a spatial classifier based on an artificial neural network (ANN), which takes spatially-related features (e.g., the density of POIs and length of highways) as input to model the spatial correlation between air qualities of different locations. The other is a temporal classifier based on a linear-chain conditional random field (CRF), involving temporally-related features (e.g., traffic and meteorology) to model the temporal dependency of air quality in a location. Read the related publications for more details.

The research has been publicly available through a "cloud + client" framework, where the cloud continuously collect real-time data, such meteorological data and air quality data. A user can access the air quality information through using a mobile client or web client.

  • A public website is:
  • A Windows Phone application, titled Urban Air, is ready for download. Click here or scan the 2D bar code on the right.
  • A Dataset is released for research purposes: download the data

    Urban Air



Step 2:

The second step is to predict the fine-grained air quality a few hours ahead. Specifically, for each monitoring station, we predict the probability of air quality status being good, normal and bad in the next a few hours. We also predict the trend of air quality being the same as current, or becoming worse, or better.

Future Work

1) Suggesting the locations for building additional monitoring stations;

2) Identifying the root cause of air pollutions;

3) Study the impact of air pollution to people's life and health.



There are a few interns who have worked with us in the urban air project. We may not be able to list all of them here.

Yubiao Chen, Xuxu Chen, Hsun-Ping Hsieh, Furui Li, Zhenni Feng.

see the CEO face-to-face interview on urban air

A story about Urban Air has been featured by GCR news.


Qiwei Jin
Qiwei Jin

Ming Li
Ming Li

Xin Liu
Xin Liu