Increase in population leads to increase in human activity which in turn leads to increase in pollution. Not only human activities but also natural processes can both generate air pollution. According to the 2014 World Health Organization, air pollution in 2012 caused the death of around 7 million people worldwide, an estimate roughly, echoed by one from International Energy Agency. The composition of normal dry air should be acknowledge to be is 78.09% NO2, 20.95% of O2, 0.93% of Ar,0.04% of CO2, 0.4% H2O and other gases. But polluted air consists of composition of particulates like NH3, CO2, SO2, NO2, VOC, CH4, PM2, PM10, toxic metals traces like lead, mercury, etc. producing toxic or poisonous gases. These gases are harmful for living begins. Harmful compositions are considered as data points. The pollutant or data points gathered from the atmospheric air data are somewhat non-linear or fuzzy. If number of unwanted composition or gases appears in atmosphere, then those are the outliers in the normal dry air. To remove outlier, data points needs to be clustered so as to group desired data. This paper gives a brief study of a fuzzy clustering concept for measuring pollution index P_i.