Dynamic Ensemble Multivariate Time Series Forecasting Model for PM2.5

In forecasting real time environmental factors, large data is needed to analyse the pattern behind values. Air pollution a major threat towards developing countries and it proliferating every year. Many methods in series prediction deep learning models estimate severity of air pollution. Each independent variable contributing necessary trend that particular locality. This approach selects multivariate coalesce updatable autoregressive model forecast Particulate matter (PM) PM2.5. To perform experimental analysis from Central Pollution Control Board (CPCB) used. Prediction carried out for Chennai with seven locations estimated PM’s using weighted ensemble method. Proposed method unveiled effective moored performance long term prediction. Dynamic budge high k-models are used simultaneously devising an helps achieve stable forecasting. Computational decreases parallel processing each sub model. Weighted shows when compared traditional like Vector Auto-Regression (VAR), Autoregressive Integrated Moving Average (ARIMA), Extended terms (ARMEX). Evaluation metrics Root Mean Square Error (RMSE), Absolute (MAE) compared.