Basics: Time Series Series Part 2
“The long run is a misleading guide to current affairs. In the long run we are all dead.” ― John Maynard Keynes, A Tract on Monetary Reform
Since one premise of our blog is that we will not be dead soon, our focus is on time series methods for “short-run” forecasting, the most common type of forecasting for business.
In Part 2 titled "Some Basics", of his series on the KDD Analytics blog, our CAO Kevin Duffy-Deno provides an overview and some definitions so we better understand how time series fits within the field of forecasting. He explains that a successful forecasting model will account for each of 3 components that may exist in a time series: trend, seasonality and cycles.
Trend - a long-run phenomenon and reflects, in business, “slowly evolving preferences, technologies, institutions and demographics.” (Diebold, Elements of Forecasting). Trends are deterministic and stochastic.
Seasonality - seasonal pattern repeats with calendar regularity which could be annual (Black Friday), quarterly or monthly, or all of these together.
Cycles - a cyclic component can also be present. Cycles are much less rigid than seasonal patterns. One example is the business cycle, from a recession low to an expansion high.
Methods for forecasting He describes numerous methods for forecasting a time series, ranging from simple to complex. The simplest is some type of smoothing routine, like moving averages or exponential smoothing (ex. is a 200-day moving average). More complex econometric methods are used to model the US economy and to generate long-run forecasts of macroeconomic variables such as GDP and employment.
More commonly used for business, in between the simpler and more complex methods, are what Kevin refers to as “time series methods.” These methods primarily rely on (but not always) the series’ historical behavior to inform the future.
A distinguishing feature of time series methods is that they explicitly account for the key characteristics of a time series: trend, seasonality and cycles. The workhorses of time series methods are single equation, least squares regression and ARIMA models.
Short Run Focus
Since most business forecasts will most likely revert to the underlying trend in the series, the best use of these time series methods is for “short-run” forecasts. He generally defines the “short run” as the period of time that matches most business’ forecast needs and will be used for planning and the issuance of guidance to investors.
Head over to Kevin's blog for the full version of Part 2, "Some Basics". We hope you find this useful in your own forecasting, but also to decode some of the jargon used by statisticians and data scientists when presenting their work.
Part 3 coming up this weekend. I'm sure the suspense is killing you.