Time series evaluation metrics in ML Python - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

Understanding Mean Absolute Error (MAE) in Time Series

Which statement best describes the Mean Absolute Error (MAE) when used to evaluate time series predictions?

A MAE calculates the average of the absolute differences between predicted and actual values, treating all errors equally regardless of direction.

B MAE measures the average of the squared differences between predicted and actual values, penalizing larger errors more heavily.

C MAE measures the correlation between predicted and actual values to assess prediction accuracy.

D MAE is the percentage of predictions that exactly match the actual values in the time series.

💡 Hint

Think about how MAE treats errors without squaring them.

MAE calculates the average absolute difference between predicted and actual values, giving equal weight to all errors regardless of their sign.

❓ Predict Output

intermediate

Calculate RMSE for a Time Series Prediction

What is the output of the following Python code that calculates the Root Mean Squared Error (RMSE) for given true and predicted values?

ML Python

import numpy as np
true = np.array([3, -0.5, 2, 7])
pred = np.array([2.5, 0.0, 2, 8])
rmse = np.sqrt(np.mean((true - pred) ** 2))
print(round(rmse, 3))

A 0.750

B 1.118

C 0.612

D 0.957

💡 Hint

Calculate squared errors, average them, then take the square root.

Squared errors: (0.5², (-0.5)², 0², (-1)²) = (0.25, 0.25, 0, 1). Mean = 1.5/4 = 0.375. RMSE = √0.375 ≈ 0.612 (option C).

❓ Model Choice

advanced

Choosing the Best Metric for Skewed Time Series Data

You have a time series dataset with many small values and a few very large spikes. Which evaluation metric is best to use to avoid the large spikes dominating the error measurement?

A R-squared (Coefficient of Determination)

B Root Mean Squared Error (RMSE)

C Mean Squared Error (MSE)

D Mean Absolute Error (MAE)

💡 Hint

Consider which metric is less sensitive to large errors.

MAE treats all errors equally by using absolute values, so large spikes do not disproportionately affect the metric, unlike MSE or RMSE which square errors.

❓ Metrics

advanced

Interpreting Mean Absolute Percentage Error (MAPE)

Given the true values [100, 200, 300] and predicted values [110, 190, 310], what is the Mean Absolute Percentage Error (MAPE) expressed as a percentage?

ML Python

true = [100, 200, 300]
pred = [110, 190, 310]
errors = [abs(t - p) / t for t, p in zip(true, pred)]
mape = sum(errors) / len(errors) * 100
print(round(mape, 2))

A 8.33

B 6.11

C 7.14

D 5.56

💡 Hint

Calculate absolute percentage errors for each point, then average.

Errors: |100-110|/100=0.1 (10%), |200-190|/200=0.05 (5%), |300-310|/300≈0.0333 (3.33%). Average ≈0.0611 or 6.11% (option B).

🔧 Debug

expert

Identify the Error in Time Series Metric Calculation Code

What error does the following Python code raise when calculating Mean Absolute Error (MAE) for time series predictions?

ML Python

true = [1, 2, 3, 4]
pred = [1, 2, 3]
mae = sum(abs(t - p) for t, p in zip(true, pred)) / len(true)
print(mae)

A No error; output is 0.0

B No error; output is 0.25

C IndexError because lists have different lengths

D ZeroDivisionError because denominator is zero

💡 Hint

Check how zip works with lists of different lengths.

zip stops at the shortest list (pred), so only pairs (1,1), (2,2), (3,3) with errors 0,0,0. Sum=0 / len(true)=4 = 0.0. No error.