Time Series Analysis

A time series can involve just one variable or several. A univariate time series tracks a single variable over time — a temperature sensor logging one reading every second is a simple example. A multivariate time series tracks several variables together and how they move relative to one another over the same period.

Whichever kind it is, the pattern a time series traces out is generally understood as a combination of four components. The secular trend (or simply trend) is the smooth, long-term direction the series moves in over an extended period — rising annual electricity consumption over a decade is a typical example, and this movement can be linear or non-linear. The seasonal component captures a regular, repeating pattern within each one-year cycle, usually driven by climate, festivals, or habits — electricity demand peaking every summer, for instance. The cyclical component is a longer, wave-like oscillation whose period stretches beyond a single year, such as the recurring booms and slumps seen in GDP data. The **irregu …