Time-Series-Forecasting--Airline-Passengers-in-Python

We will use Box-Jenkins methodology for time-series analysis :

1. Data conditioning & Model Selection : TS decomposition and Making TS stationary (Differencing & Dikey Fuller Test)
2. Estimation of model parameters : Compute coefficients of model
3. Model Validation : Analyzing residuals (normality, white noise, Ljung-Box test)

We used different ARIMA models to do the forecasting and compared them based on evaluation metrics like AIC,BIC and Log-likelihood values. Finally we selected the best model which was fulfilling all the criteria and statistical tests.