Work in Progress

1. Probability of more extended forecasting and improved accuracy over training three years of data in comparison to one year's data using Vector AutoRegression
2. Finer Forecasting of Neural Network's model over Vector AutoRegression

1. Available: Pollutant concentration levels with a frequency of 15 min
2. Additional (if possible): Wind speed, rainfall, temp, humidity, pressure

1. Handling Null data
2. Upsampling weather data to merge with pollutant data as there will be a difference in the frequency of data recorded

Various neural networks models for multivariate time series prediction

1. Season (spring, summer, monsoon, winter)
2. Morning, Afternoon, Evening, Night

Include the Topography feature, and train the model, along with more than one place.

1. Data Collection
2. Data Stationarity Check
3. Data Processing (Missing values imputation)
4. Exploratory Data Analysis (Correlation, Autocorrelation, ADF test, Granger-causality Test)
5. Train models
6. Forecast
7. Model Evaluation