At present time, the forecasting of air particles has been a current research topic due to increase in bad air quality because of industrialization and increase in pollution due to vehicle and in COVID-19 when there was lockdown it came in our notice that this thing can be controlled its effects which was highlighted due to COVID 19 lockdown period. It includes a various source of contamination, making it hard to decide the entirety of the contributing meteorological and ecological components. At the point when just the PM 2.5 fixationtime sequences are taken without other external data, precise estimating is significant and productive. For address, this issue this paper presents the ARIMA based model for forecast PM 2.5 data concerning time. In this paper, two methods are proposed. First dividing the data (80 % training) and (20 % testing) then decomposes one-dimensional data through wavelet decompositionof level-2 dB2. Then, it uses ARIMA model method to forecast each divided sequel and reconstruct its predicted results to obtainthe finalize predicting outcomes. Second without decomposes the data, we directly apply the ARIMA model and forecast the results. The ARIMA model has forecast more accurate results concerning predicting the concentration of PM 2.5 as compare to the WAVELET-ARIMA model. The two proposed ARIMA and Wavelet ARIMA can be efficiently applied to forecastingPM 2.5 concentration in short term and can be enhancing the accuracy. Moreover, relating the forecasted results with the policygoverning to control the pollution as shown by implementinglockdown as PM 2.5 value has been reduced up to 50% in different cities during the lock down period which can be seen from study.

Keywords : Particulate Matter 2.5, Auto Regressive Integrated Moving Average, Mean Absolute Value, Weather Research and Forecasting, Carbon Matter, Elemental Carbon, Root Mean Square Error.