Stationarity and differencing in ML Python - Cheat Sheet & Quick Revision

Recall & Review

beginner

What does it mean for a time series to be stationary?

A stationary time series has constant mean, constant variance, and constant autocovariance over time. This means its behavior does not change as time passes.

beginner

Why is stationarity important in time series analysis?

Many forecasting models assume stationarity because it makes patterns stable and predictable. Without stationarity, models may give unreliable predictions.

beginner

What is differencing in time series?

Differencing means subtracting the previous value from the current value to remove trends or seasonality, helping to make a time series stationary.

beginner

How do you perform first-order differencing on a time series?

First-order differencing subtracts each value by the value immediately before it: new_value = current_value - previous_value.

intermediate

What is the effect of over-differencing a time series?

Over-differencing can remove important information and add unnecessary noise, making the series harder to model and interpret.

Which of the following is NOT a characteristic of a stationary time series?

A Constant autocorrelation structure

B Constant variance over time

C Constant mean over time

D Increasing trend over time

A stationary time series does not have trends like increasing or decreasing patterns.

What is the main purpose of differencing a time series?

A To make the series stationary

B To add noise to the series

C To smooth the series

D To increase the series length

Differencing helps remove trends or seasonality to achieve stationarity.

How is first-order differencing calculated?

A Subtracting the previous value from the current value

B Adding the current and previous values

C Subtracting the current value from the next value

D Dividing the current value by the previous value

First-order differencing subtracts the previous value from the current value.

If a time series is already stationary, what happens if you apply differencing?

A It makes the series non-stationary

B It improves the model accuracy

C It may add unnecessary noise

D It has no effect

Differencing a stationary series can add noise and remove useful information.

Which test is commonly used to check stationarity?

A T-test

B Augmented Dickey-Fuller test

C Chi-square test

D ANOVA

The Augmented Dickey-Fuller test checks if a time series is stationary.

Explain in your own words what stationarity means and why it matters in time series forecasting.

Think about how a series that changes its behavior over time might confuse a forecasting model.

Describe how differencing helps to make a time series stationary and what risks come with over-differencing.

Consider differencing as a way to flatten the series but too much can harm it.