Let's try to understand how good our predictions are for each of the stocks:
- Netflix (NFLX): The following diagram illustrates the prices of the Netflix stock from 2002 through 2018:
The price for the year 2018 is defined using the two vertical lines. It shows the growth in the price of the stock throughout the entire year.
As per the first problem case, we consider the period from 2008-2016 for training:
Normalizing the prices by each year for modeling gives us the following plot:
Predicting the prices of the stock for the whole year of 2017 with a 95% confidence interval gives us the following plot:
Comparing of the generated values with the actual values, it is clear that the model falters by predicting the value to be less than the actual value. However, the reason for this could be the highs and lows of the prices of Netflix in the year 2016. These are not captured through the basic-level kernel used in this project.
For the second problem case, we consider...