Chapter 6: Calculating Forecasts

Study Guide Highlights

Trend and seasonal patterns

Time-series analysis separates recurring patterns from the underlying direction of the data. Trend and seasonal variation receive particular attention, along with the role of moving averages in smoothing short-term fluctuations so that the trend is easier to see. Additive seasonal effects, expressed as amounts, are distinguished from multiplicative effects, expressed proportionately, because the form of the data determines the appropriate interpretation.

Regression and linear relationships

Simple regression is used to describe a broadly linear relationship between variables and to support estimation where that relationship is sufficiently stable. A line inferred from a wider data set is distinguished from a line derived from only two points. Forecast use therefore depends not only on obtaining a line, but also on considering whether the data pattern and surrounding conditions make that line credible.

Index numbers and expected values

Index numbers provide a common base for comparing changes over time and can be used to adjust or interpret monetary data. The concept of an expected value as a probability-weighted outcome is also introduced. Both techniques require careful interpretation: an index depends on its base and comparison point, while an expected value summarises uncertainty rather than describing an outcome that must actually occur.

Sensitivity and forecast reliability

Forecasts can be useful without being certain. The chapter considers the dependence of forecasting methods on historical patterns, data quality and stable relationships, then uses sensitivity analysis to show how changes in important assumptions affect the result. The emphasis is on identifying material exposure and linking a recommendation to the evidence, rather than treating a single base-case forecast as a guaranteed future outcome.