I am trying to learn time series prediction and forecasting in Python. I have plotted the ACF and PACF of my total electron content which has a seasonality i.e. TEC value gets maximum at day time and min at night time. Overall the data has no upward or downward trend and the test statistics from Adfuller test is -3.67
I've got the following graphs where ACF is tailing off but so is the PACF and now I am confused about which would be the best coefficients for ARIMA model.
ACF and PACF plots of Timeseries TEC
NOTE: I want to forcast 10 days after and 20 days before earthquake and then compare it with the actual values to get a differenced value and show the impact of earthquake on total electron content.
The TEC values are also affected by geomagnetic storm so next I will train a machine learning model to classify the impact of earthquake and impact of space weather.
I can share my data if anyone wants to see it.
Thank you!
I am trying to fit an ARIMA model to forcast timeseries TEC values 20 days before and 10 days after the earthquake. My goal is to get a forcast/prediction for a specific timerange and then compare it with the actual value to see how much difference is there due to earthquake.
I am struck at selecting the AR and MA coefficients for the data.