Transformers and time series modeling

Before jumping into transformers for time series modeling, it is helpful to understand basic approaches such as the moving average. This approach takes a window of historical data and applies an average to find the future value. Then, it continues to iterate over the data and find the next value using the next steps:

1. In order to illustrate this approach, we first need to get the data using yfinance, and then install the package:

   pip install yfinance

2. Then, you can easily fetch the data:

   import yfinance as yf
   btc_ticker = yf.Ticker("BTC-USD")
   btc_data = btc_ticker.history(period="max")

   The data will be fetched, and you can easily see it by printing out the variable:

   btc_data

   It will show you a nice pandas DataFrame, as shown in Figure 18.2.

Figure 18.2 – Historical Bitcoin data

3. You can also easily visualize it with the following code:

   from matplotlib import pyplot as plt
   plt.plot...