Stationarity and differencing in ML Python - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

Why is stationarity important in time series analysis?

Which of the following best explains why stationarity is important when analyzing time series data?

A Stationarity guarantees the time series has a linear trend that can be easily predicted.

B Stationarity means the data has no missing values, which is essential for model training.

C Stationarity ensures the statistical properties like mean and variance do not change over time, making modeling and forecasting more reliable.

D Stationarity means the time series data is always increasing, which simplifies analysis.

💡 Hint

Think about why consistent behavior in data helps in making predictions.

Stationarity means the data's mean, variance, and other properties stay constant over time. This consistency helps models learn patterns that hold in the future.

❓ Predict Output

intermediate

Output of differencing a time series

Given the time series data [3, 5, 8, 12, 18], what is the result of applying first-order differencing?

ML Python

data = [3, 5, 8, 12, 18]
differenced = [data[i] - data[i-1] for i in range(1, len(data))]
print(differenced)

A [1, 2, 3, 4]

B [3, 5, 8, 12, 18]

C [5, 8, 12, 18]

D [2, 3, 4, 6]

💡 Hint

Subtract each value from the previous one.

First-order differencing subtracts each value by the one before it: 5-3=2, 8-5=3, 12-8=4, 18-12=6.

❓ Model Choice

advanced

Choosing a model for non-stationary data

You have a time series with a clear upward trend and non-constant variance. Which model approach is best suited before forecasting?

A Apply differencing to make the series stationary, then use an ARIMA model.

B Use a linear regression model directly on the raw data without transformation.

C Apply PCA to reduce dimensionality before forecasting.

D Use K-means clustering to group similar time points.

💡 Hint

Think about how to handle trends and changing variance before modeling.

ARIMA models require stationarity. Differencing removes trends and stabilizes variance, making ARIMA suitable.

❓ Metrics

advanced

Evaluating stationarity with statistical tests

Which metric or test is commonly used to check if a time series is stationary?

A Augmented Dickey-Fuller (ADF) test

B Confusion Matrix

C Mean Squared Error (MSE)

D Silhouette Score

💡 Hint

Look for a test that checks for unit roots in time series.

The ADF test checks if a time series has a unit root, indicating non-stationarity. A low p-value suggests stationarity.

🔧 Debug

expert

Identifying the error in differencing code

What error will this code produce when trying to difference a time series?

data = [10, 15, 20]
diff = [data[i] - data[i+1] for i in range(len(data)-1)]
print(diff)

A IndexError because it tries to access an element outside the list range.

B Produces incorrect differencing values because it subtracts the next value from the current instead of the current from the previous.

C SyntaxError due to incorrect list comprehension syntax.

D Runs correctly and outputs [5, 5].

💡 Hint

Check the order of subtraction and the indices used.

The code subtracts the next value from the current, which reverses the differencing direction. It does not cause an index error because the range is correct, but the values are wrong.