2.1 Components of Time Series

Time series data is like a puzzle with four main pieces: trend, seasonality, cycles, and random stuff. Knowing these parts helps us predict what might happen next. It's like figuring out if a business is growing, when it's busiest, and what surprises might pop up.

Breaking down time series data is super useful. It's like taking apart a watch to see how it works. We can spot patterns, make better guesses about the future, and understand why things happen. It's a key skill for anyone working with data over time.

Time series components

Definition and importance

• A time series is a sequence of data points collected and recorded at regular intervals over time (hourly, daily, monthly, or yearly)
• The four primary components that make up a time series are trend, seasonality, cyclical, and irregular (residual) components
• Each component contributes to the overall pattern and behavior of the time series
• Understanding these components is crucial for accurate forecasting and analysis

Decomposition process

• Decomposition is the process of separating a time series into its individual components to better understand the underlying patterns and make more accurate predictions
• Additive decomposition assumes that the components of the time series are independent and can be added together to form the original data, using the formula: $\text{Time Series} = \text{Trend} + \text{Seasonality} + \text{Cyclical} + \text{Irregular}$
• Multiplicative decomposition assumes that the components of the time series interact with each other and can be multiplied to form the original data, using the formula: $\text{Time Series} = \text{Trend} \times \text{Seasonality} \times \text{Cyclical} \times \text{Irregular}$

Trend vs seasonality vs cyclical vs irregular

Trend component

• Trend refers to the long-term direction or general tendency of a time series, which can be increasing, decreasing, or stable over time
• Represents the underlying growth or decline in the data
• Provides information about the long-term direction and rate of change in the time series
• Helps identify whether the data is generally increasing, decreasing, or remaining stable over time

Seasonality component

• Seasonality describes regular, predictable fluctuations that occur within a fixed period (year or month) and repeat in a similar pattern
• Captures the regular, periodic fluctuations within the time series
• Allows for the identification of patterns that repeat at fixed intervals (yearly or monthly cycles)
• Examples include increased retail sales during the holiday season or higher ice cream sales in summer months

Cyclical component

• Cyclical components are patterns of alternating periods of expansion and contraction that are not of fixed duration
• Usually influenced by external factors (economic conditions or business cycles)
• These fluctuations typically last longer than a year
• Represent longer-term oscillations that are not of fixed duration
• Provide insight into the impact of external factors on the time series (economic expansions and contractions)

Irregular component

• Irregular or residual components are random, unpredictable fluctuations that are not captured by the other components
• Represent the remaining variation after accounting for trend, seasonality, and cyclical effects
• Can be caused by one-off events, measurement errors, or other unforeseen factors
• Account for the remaining, unpredictable variations in the time series after the other components have been identified
• Help isolate the impact of random events or noise in the data

Component contributions to time series patterns

Interaction of components

• The interaction and combination of these components create the unique pattern and behavior of a given time series
• Understanding their individual contributions is essential for accurate forecasting and decision-making

Moving averages and smoothing techniques

• Moving averages and smoothing techniques (simple moving average or exponential smoothing) can be used to estimate and remove the trend and cyclical components from the time series

Seasonal decomposition methods

• Seasonal decomposition methods (classical seasonal decomposition or STL - Seasonal and Trend decomposition using Loess) can be employed to identify and extract the seasonal component from the time series

Time series data decomposition

Residual analysis

• Residual analysis involves examining the remaining irregular component after the trend, seasonality, and cyclical components have been removed
• Assesses the model's goodness of fit and identifies any patterns or outliers in the residuals

Benefits of decomposition

• Decomposition helps to better understand the underlying patterns and drivers of the time series
• Enables more accurate forecasting, anomaly detection, and decision-making based on the individual components
• Separating the original time series data into its trend, seasonality, cyclical, and irregular components uses various mathematical and statistical techniques