In the world of data science and statistics, most datasets are like a still photograph—a snapshot of a moment in time. However, Time Series Analysis is more like a motion picture. It is the study of data points collected or recorded at specific time intervals, where the order of the data matters just as much as the values themselves. Whether you are forecasting the next peak in Kenya’s inflation rate, predicting stock market volatility, or analyzing seasonal sales for a retail business, you are leaning on the rigorous framework of time-dependent modeling.
Below is the exam paper download link
PDF Past Paper On Time Series Analysis For Revision
Above is the exam paper download link
To help you move from simply looking at “old data” to accurately predicting “future trends,” we have curated a deep-dive Q&A session based on high-frequency exam topics.
What is the fundamental difference between ‘Stationary’ and ‘Non-Stationary’ Data?
This is the “Golden Rule” of Time Series.
-
Stationary Data: The statistical properties (mean, variance, and autocorrelation) are constant over time. It’s like a heartbeat that stays steady; it’s predictable and easy to model.
-
Non-Stationary Data: This data has a trend (it’s going up or down) or seasonality (it repeats patterns).
In an exam, you will likely be asked to turn non-stationary data into stationary data using Differencing. This involves subtracting the previous value from the current one to “flatten” the trend.
How do we define ‘Seasonality’ versus ‘Cyclical’ patterns?
While they sound similar, they are mathematically distinct:
-
Seasonality: These are fluctuations that repeat at fixed, regular intervals (e.g., increased ice cream sales every December or higher electricity usage every evening).
-
Cyclical: These are fluctuations that don’t have a fixed period and are often linked to longer-term economic shifts (like a 5-year business cycle).
A common past paper challenge involves using a Seasonal Decomposition to strip these patterns away and see the underlying “Noise” and “Trend.”
What are ‘Autocorrelation’ (ACF) and ‘Partial Autocorrelation’ (PACF)?
These are the “Fingerprints” of your data.
-
ACF: Measures the correlation between a series and its lagged values (e.g., how today relates to yesterday, the day before, and so on).
-
PACF: Measures the correlation between today and a specific lag (like 2 days ago) after removing the influence of the days in between.
You use these plots to decide whether to use an AR (Autoregressive) or MA (Moving Average) model.
What is the ‘ARIMA’ Model?
ARIMA stands for AutoRegressive Integrated Moving Average. It is the “Swiss Army Knife” of forecasting.
-
AR (p): Uses the relationship between an observation and a number of lagged observations.
-
I (d): The number of times the data was differenced to make it stationary.
-
MA (q): Uses the dependency between an observation and a residual error from a moving average model applied to lagged observations.
If you can correctly identify the $(p, d, q)$ parameters in an exam sitting, you are well on your way to an “A.”
How do we test for ‘Random Walks’ and ‘Unit Roots’?
A “Random Walk” is the nightmare of a forecaster—it means the data is moving purely by chance and the past cannot predict the future. To check for this, we use the Augmented Dickey-Fuller (ADF) Test. If the test shows a “Unit Root,” your data is non-stationary and needs serious differencing before you can trust your forecast.
What is ‘White Noise’ in a Time Series?
If your model is perfect, the “Residuals” (the errors left over) should look like White Noise. This means the errors have a mean of zero, constant variance, and no correlation. If your residuals still show a pattern, it means your model missed some important information, and you need to go back to the drawing board.
Conclusion
Time Series Analysis is as much an art as it is a science. It requires you to look past the chaos of daily fluctuations to find the rhythmic heartbeat of the data. Success in your finals comes from your ability to diagnose a series—identifying whether it needs a simple linear trend or a complex SARIMA (Seasonal ARIMA) approach.
To help you practice your lag calculations and master the ADF test, we have provided a link to a comprehensive PDF revision resource below.
Last updated on: March 25, 2026