How Time Series Forecasting is Different from a Typical Regression Task
And how to prepare a time series for regression using time-delay embedding
“A time series is a set of data points ordered in time.”
That is how we define a time series. In a time series, data is collected sequentially over time, usually at regular intervals (e.g., daily, monthly, yearly).
“Forecasting is predicting the future using historical data.”
On the other hand, “Regression is predicting a numerical value (real number) using the values of input variables.”
Both time series forecasting and regression are used to predict future values.
Time series forecasting seems like a typical regression task, but it doesn't!
There are some key differences between time series forecasting and typical regression. Here, typical regression means regression with time-independent scenarios.
Time Series Forecasting vs Regression: Key Differences
1. Nature of data
Time series have an order, but regression tasks are performed with data that does not have an order. In a time series, the data points are indexed by time…
