Different damages and air pollutions caused by chemical gas leaks from industrial sites and automotive exhausts, demand for gas monitoring method has recently increased. Because of the production of toxic gases by industry, vehicular emissions, and higher concentrations of dangerous gases as well as some matter in environment, air is becoming increasingly polluted. This research proposed novel technique in air quality analysis by environmental gas sensing by semi-conductor type gas sensor using micro electromechanical systems (MEMS). The monitored data has been collected based on IoT module and processed for classifying the air quality based on monitored environment data. The classification of monitored data has been carried out using multilayer convolutional perceptron neural network (MCPNN). Based on a sensor simulation model, we offer a characterisation scheme for analysing the resilience of several machine learning models for ambient gas sensing. Behaviour of semiconductors, heights of surface barrier potentials, and gas sensing properties of sensing layers has all been studied. Interactions of humidity with semiconductor oxides are studied in order to correct for its effect on gas detection using an algorithm.