In this notebook, we will explore two time series models to fit our current daily price data and make predictions based on our fitted models. The first method is Auto-ARIMA model, the second one is SARIMAX model.

The advantage of the first model is that it can provide a stationary series using ADF test, the disadvantage is complicated data manipulation, for example: taking difference, cut-off partial series etc.

The advantage of the second mode is that it can deal with the original dataset and capture the trend and seasonality directly. Statistical results show that the second model is more suitable to forecast the current dataset and can be easily generalized to a larger dataset.