Time-Based Data

The One Thing

Time-based data analysis examines observations in chronological order to identify trend, variation, and change over time. Percentage changes, moving averages, index numbers, graphs, and forecasting methods are used to compare observations, smooth fluctuations, and make informed estimates about future behaviour.

Definitions and Results

• Time Series: A set of observations recorded in time order, such as monthly sales, yearly population, or daily temperature.
• Time Variable: The variable identifying when an observation was recorded, such as year, quarter, month, week, or day.
• Chronological Order: The arrangement of data from the earliest time period to the latest time period. Time-based data should be arranged this way before analysis.
• Trend: The long-term general direction of movement, which may be upward, downward, or approximately constant.
• Seasonal Variation: A regular, repeating pattern occurring within a year, such as higher ice-cream sales during summer. Seasonal patterns should be compared with the same season in another year, such as January with January, rather than only with the immediately preceding month.
• Cyclical Variation: Long-term rises and falls associated with business or economic cycles, usually lasting longer than one year.
• Irregular Variation: Unexpected changes caused by unusual events such as floods, strikes, pandemics, or sudden policy changes.
• Level: The typical or average value around which observations are located during a particular period.
• Absolute Change:

$$\text{Absolute Change}=\text{Later Value}-\text{Earlier Value}.$$

• Percentage Change:

$$\text{Percentage Change}=\frac{\text{Later Value}-\text{Earlier Value}}{\text{Earlier Value}}\times 100.$$

• Percentage Increase:

$$\frac{\text{Increase}}{\text{Original Value}}\times 100.$$

• Percentage Decrease:

$$\frac{\text{Decrease}}{\text{Original Value}}\times 100.$$

• Moving Average: An average calculated over a fixed number of consecutive observations and moved through the series to smooth short-term fluctuations.
• Simple Moving Average: For $k$ observations,

$$\text{Moving Average}=\frac{\text{Sum of the }k\text{ observations}}{k}.$$

• Weighted Moving Average: An average in which different observations receive different weights. The weights usually add to $1$ or $100\%$.
• Index Number: A statistical measure showing the relative change in a variable compared with a selected base period.
• Base Period: The reference period whose value is generally taken as $100$.
• Simple Price Index:

$$\text{Index}=\frac{\text{Current Period Price}}{\text{Base Period Price}}\times 100.$$ An index of $100$ indicates no change from the base period; an index above $100$ indicates an increase; an index below $100$ indicates a decrease.

• Forecasting: Estimating future values using past and present time-based data and identified patterns.
• Interpolation: Estimating a value within the range of known observations.
• Extrapolation: Estimating a value beyond the range of known observations, such as predicting a future value.
• Line Graph: A graph joining time-ordered observations to show direction, fluctuations, and turning points.
• Linear Trend Model:

$$y=a+bt,$$ where $y$ is the estimated value, $t$ represents time, $a$ is the intercept, and $b$ is the average change per time unit. A positive slope indicates an overall increase, while a negative slope indicates an overall decrease.

Worked Methods

1. Organising a Time Series

• Identify the time variable: year, quarter, month, week, or day.
• Arrange the observations in chronological order, from the earliest period to the latest.
• Identify the time interval and keep it consistent where possible.
• Inspect the data for trend, seasonal variation, cyclical variation, irregular variation, and level.

The order of observations matters because changing the time sequence can hide or distort trends and patterns. For example, monthly sales should be listed from the earliest month to the latest month rather than rearranged by size.

2. Calculating Absolute and Percentage Change

• Identify the earlier value and the later value.
• Calculate the absolute change:

$$\text{Later Value}-\text{Earlier Value}.$$

• For a percentage change, divide the absolute change by the earlier value and multiply by $100$:

$$\frac{\text{Later Value}-\text{Earlier Value}}{\text{Earlier Value}}\times 100.$$

• Use the original value as the denominator for both percentage increase and percentage decrease.
• Interpret the sign: a positive result indicates an increase and a negative result indicates a decrease.

Percentage changes are often more meaningful than absolute changes when comparing quantities with different sizes. If the base-period value is zero, percentage change cannot be calculated directly.

3. Calculating a Simple Moving Average

• Select $k$ consecutive observations.
• Add the $k$ values.
• Divide the sum by $k$:

$$\text{Moving Average}=\frac{\text{Sum of values for }k\text{ consecutive periods}}{k}.$$

• Move the group forward by one period.
• Repeat the calculation through the series.

A moving average smooths random fluctuations and helps reveal underlying movement. However, it reduces the number of usable observations at the ends of the series and may conceal important sudden events.

For an even number of periods, such as a 4-quarter moving average, centring may be required to place the averages correctly against the time periods.

4. Using a Weighted Moving Average

• Select the observations to be included.
• Assign each observation its specified weight.
• Multiply each observation by its weight.
• Add the weighted values.
• Check that the weights add to $1$ or $100\%$.
• Use the total as the weighted moving average.

A weighted moving average gives greater influence to selected observations, usually the more recent ones. Like a simple moving average, it smooths fluctuations but does not explain the causes of changes.

5. Constructing and Interpreting a Simple Price Index

• Select a meaningful and reliable base period, preferably one free from unusual disturbances.
• Take the base-period value as $100$.
• Substitute the current-period price and base-period price into:

$$\text{Index}=\frac{\text{Current Period Price}}{\text{Base Period Price}}\times 100.$$

• Interpret the result:

- $100$: no change from the base period; - above $100$: an increase; - below $100$: a decrease.

If the base-period value is zero, an ordinary index calculation cannot be used directly. Index numbers simplify comparison across time by expressing values relative to a common base period.

6. Representing and Interpreting a Time Series Graphically

• Put time on the horizontal axis.
• Put the measured variable on the vertical axis.
• Use a suitable time scale and appropriate units.
• Plot the observations in chronological order.
• Join the observations to form a line graph.
• Include a suitable title, labelled axes, appropriate units, and a clear time scale.
• Examine the graph for trend, fluctuations, and turning points.

A line graph makes changes over time easier to interpret and communicate. Long-term trend and short-term fluctuations should be separated when the aim is to understand underlying movement.

7. Estimating a Trend and Forecasting

• Use suitable historical data recorded in chronological order.
• Identify the general direction of movement.
• Fit or interpret a trend model, such as:

$$y=a+bt.$$

• Interpret $b$:

- $b>0$: overall increase; - $b<0$: overall decrease.

• For interpolation, estimate a value within the range of known observations.
• For extrapolation, use the model to estimate a value beyond the known range, such as a future value.
• State that the result is an estimate and consider uncertainty.

Forecasts may be affected by structural changes, unusual events, poor-quality data, or a trend that does not continue. Correlation between time and a variable does not, by itself, prove that time caused the change.

Where It Goes Wrong

• Failing to arrange observations in chronological order can hide or distort trends and patterns.
• Omitting or changing the time interval makes comparisons unreliable; the periods should be clearly identified and kept consistent where possible.
• Using the wrong denominator for percentage change: the earlier or original value must be used, not the later value.
• Treating a moving average as an explanation of change is incorrect; it smooths short-term fluctuations but cannot identify their causes.
• Forgetting to centre a moving average when using an even number of periods, such as a 4-quarter moving average, places the averages against the wrong time periods.
• Interpreting a forecast as certain overlooks structural changes, unusual events, poor-quality data, and the possibility that the trend will not continue.

What Gets Asked

This material supports questions requiring students to:

• Define a time series, time variable, chronological order, trend, seasonal variation, cyclical variation, irregular variation, level, interpolation, extrapolation, forecasting, and index numbers.
• Arrange observations in chronological order and identify the relevant time intervals.
• Calculate absolute change, percentage change, percentage increase, and percentage decrease.
• Calculate and interpret simple moving averages, including a 4-quarter moving average requiring centring.
• Calculate weighted moving averages using specified weights.
• Construct and interpret a simple price index using a selected base period.
• Explain the meaning of index values of $100$, above $100$, and below $100$.
• Read or draw a line graph with a suitable title, labelled axes, units, and time scale.
• Identify trend, seasonal, cyclical, and irregular variation from a time series.
• Interpret the slope and parameters of the linear trend model $y=a+bt$.
• Distinguish between interpolation and extrapolation.
• Explain the limitations of moving averages, forecasts, and conclusions based on correlation between time and a variable.