After seeing the first week’s inctoduction videos, I’m not exactly seeing what’s the main difference between time series and regression task.
In other words, if we can approach time series with regression for prediction, from where does this this term appear and how does it differ from the other task?
Thanks in advance.
You should be able to solve both regression and classification tasks when it comes to time-series data. For instance, if you were given weather tags like (sunny / cloudy / rainy etc) per day, your prediction might be the weather descriptions for one of more days in the future. The ones in the lectures and assignment are instances of regression problems.
When it comes to a vanilla regression problem, you don’t have to think in terms of timesteps. Each row is an observation in isolation. In time series data, order of observations matter. Factors such as trend and seasonality come into play in a time series dataset. The goal of a time series problem is to remove trend and seasonality from the data and build a model for what is left behind aka the noise.