Our research is aimed at predicting the air quality. The data is obtained from examining 31 monitoring stations which were distributed to 31 grid squares over some locations in Xiamen located in south east coast of china. 6 air pollutants i.e. Sulphur (IV) oxide, nitrogen (ii) oxide, O3, carbon (ii) oxide, TSP and dust fall (DF) were monitored by the different stations. Through stepwise discriminant analysis only SO2, NO2 and DF were selected to avoid the excessive calculation in the modeling process. The major sources of air pollution giving rise to the pollutants included industrial coal consumption, population density, traffic flow, and shopping density. There were other sources but were however rejected on various grounds such as inability to be quantified during modeling, significantly low correlation between the source of pollutant and the quantity of pollutant in the air among other reasons. The method used to analyze and develop the models is the stepwise cluster analysis. An R based statistical package called rSCA (r package for stepwise cluster analysis) was strictly used to facilitate the computation. The computation and analysis yielded the tree and map which, at an à level of significance, could predict the pollutant given a set of independent variables (source of pollution).Eventually this article will show the impact of the various sources of pollution and the pollutants to the general air quality.