Hi All,
I am looking at time series from the course. To summarise my understanding you get a time series date and data at that point in datetime. You then segment the series into training and validation. From the training you then break into features and labels for a predefined window size. You will build the model and then predict. My question why are you using the validation data to predict the accuracy of your model. Lets say I am measuring stock price, I use the last 20 days to build a model, how do I get tomorrows stock price etc? Thanks in advance.