khwajawisal/Time-series-analysis ? reverse-engineered prompt

Build me a simple time series forecasting project in notebooks that predicts Delhi air quality using historical pollution data from two monitoring stations, one for Dwarka and one for IBHAS. I want the notebooks to load the data, clean it up, prepare it for a vanilla LSTM model, train the model, and show the predicted PM2.5 or AQI values clearly against the actual values.

Please make it easy to follow, since this is meant for learning time series forecasting. Include the basic steps for preprocessing, splitting the data, scaling it, creating sequences for the LSTM, training, evaluating, and plotting the results. If you need to look up current best practices for LSTM notebook workflows, go ahead and do that. Keep the code organized so each notebook can run on its own and explain what is happening along the way.