This week, we tackle the trending issue. No, we aren’t talking about the latest trend in technology or fashion; we’re talking about trend analysis for time series data! Disappointed? Don’t be, this is an exciting and fun topic.
Put simply, trending is the practice of fitting a curve (e.g. line, polynomial, exponential, etc.) to your data over time, in an effort to project a forecast and establish a confidence interval.
How does this relate to time series analysis? While the fitting curve is a function of time, the parameters of the curve were found using prior information and are thus related.
Why should we care? The trend is very often used (or abused) in the industry to make a quick (and dirty) forecast. Executives might use the trending tool as a sanity check when he/she examines results from more advanced models.
In this paper, we will go over the “NxTrend” built-in function that was first introduced in NumXL 1.55 (LYNX) and demonstrate, through numerous examples, its use and the intuition behind it. We will focus on the backtesting aspect and the forecast confidence interval.