Time Series Components

Why This Matters

Time series analysis is the backbone of forecasting, and your ability to decompose data into its fundamental components will determine how well you can model real-world phenomena. You're being tested on your understanding of systematic patterns versus random variation, how to identify and isolate each component, and when to apply specific techniques to transform messy data into something analyzable. These concepts connect directly to regression, ARIMA modeling, and forecast evaluation—topics that build on everything covered here.

Don't just memorize definitions—know why each component matters for model building and how they interact. When you see a time series on an exam, you should immediately ask: What's the trend? Is there seasonality? Is this stationary? Master these components, and you'll have the conceptual foundation for every forecasting method that follows.

---

Systematic Patterns: The Predictable Structure

These components represent the portions of your time series that follow recognizable, modelable patterns. Systematic patterns are what forecasting models try to capture—everything else is noise.

Trend

• Long-term directional movement in the data—can be upward (growth), downward (decline), or flat (stability)
• Linear or nonlinear—trends may follow $y = \beta_0 + \beta_1 t$ or more complex polynomial/exponential forms
• Critical for forecasting because extrapolating trend is often the primary goal of time series analysis

Seasonality

• Fixed-period, recurring patterns that repeat at known intervals—daily, weekly, monthly, quarterly, or annually
• Driven by external factors like weather, holidays, school calendars, or fiscal year cycles
• Distinguishable from cycles because the period is constant and known in advance (e.g., retail sales spike every December)

Cyclical Patterns

• Long-term fluctuations tied to business or economic cycles, typically spanning multiple years
• Variable and unpredictable period—unlike seasonality, you can't say "this repeats every X months"
• Harder to model because cycles don't follow a fixed schedule and may require leading economic indicators to forecast

Level

• Baseline value around which the series fluctuates—think of it as the "center of gravity" of your data
• Reference point for decomposition—trend describes movement away from level, seasonality describes oscillation around it
• Shifts in level (structural breaks) can indicate regime changes requiring model adjustment

---

Random Variation: The Unpredictable Noise

Not everything in a time series can be explained by patterns. Understanding what's random—and confirming that your model residuals look random—is essential for valid inference.

Irregular Fluctuations

• Unpredictable, non-repeating variations caused by one-time events—natural disasters, policy shocks, or market surprises
• Also called residuals or noise after systematic components are removed
• Cannot be forecasted but must be accounted for in prediction intervals and model diagnostics

White Noise

• Purely random series with $\text{mean} = 0$, constant variance, and zero autocorrelation at all lags
• Benchmark for model adequacy—if your residuals aren't white noise, your model missed something systematic
• Defined formally as $\epsilon_t \sim WN(0, \sigma^2)$ where observations are uncorrelated

---

Statistical Properties: What Makes Analysis Possible

These concepts determine whether standard time series methods will work on your data. Stationarity and autocorrelation are diagnostic checkpoints before any serious modeling begins.

Stationarity

• Constant mean and variance over time—the statistical properties don't depend on when you observe the series
• Required for many models including ARMA—non-stationary data violates assumptions and produces spurious results
• Achieved through transformation via differencing ($y_t - y_{t-1}$), log transforms, or detrending

Autocorrelation

• Correlation of a series with its own lagged values—measures how today's value relates to yesterday's, last week's, etc.
• Quantified by ACF (autocorrelation function) showing correlation at each lag $k$: $\rho_k = \text{Corr}(y_t, y_{t-k})$
• Guides model selection—ACF and PACF plots reveal whether AR, MA, or ARMA structures are appropriate

---

Analytical Techniques: Tools for Understanding

These methods help you extract insights from raw time series data. Decomposition separates the signal; smoothing reveals it.

Decomposition

• Separates trend, seasonality, and residuals to reveal underlying structure hidden in raw data
• Additive model when components add: $y_t = T_t + S_t + \epsilon_t$; multiplicative when they interact: $y_t = T_t \times S_t \times \epsilon_t$
• Choose multiplicative when seasonal swings grow proportionally with level (e.g., retail sales with larger December spikes as baseline grows)

Moving Average

• Smoothing technique that averages consecutive observations to reduce noise and reveal trend
• Simple MA weights all $k$ observations equally: $MA_t = \frac{1}{k}\sum_{i=0}^{k-1} y_{t-i}$; weighted MA emphasizes recent values
• Tradeoff in window size—larger $k$ = smoother trend but more lag; smaller $k$ = responsive but noisier

---

Quick Reference Table

---

Self-Check Questions

1. A retail company notices sales increase every December but also sees larger overall swings during economic expansions. Which two components explain these patterns, and how do they differ in predictability?

2. You've fit a forecasting model and want to check if it captured all systematic patterns. What statistical property should the residuals exhibit, and how would you test for it?

3. Compare and contrast additive versus multiplicative decomposition. Under what data conditions would you choose one over the other?

4. Your time series has an upward trend and increasing variance over time. Which two techniques might you apply to achieve stationarity, and in what order?

5. If an FRQ asks you to "identify the appropriate model structure" for a dataset, which diagnostic tool would you use to examine dependence across time lags, and what patterns would suggest an AR versus MA component?