ARIMA model basics in ML Python - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

Understanding ARIMA Components

Which of the following correctly describes the 'I' component in an ARIMA model?

A It represents the number of times the data is differenced to make it stationary.

B It stands for the number of autoregressive terms in the model.

C It is the parameter controlling seasonal effects in the data.

D It indicates the number of moving average terms included.

💡 Hint

Think about what 'differencing' means in time series.

The 'I' in ARIMA stands for 'Integrated', which means the number of times the data is differenced to remove trends and make it stationary.

❓ Predict Output

intermediate

ARIMA Model Prediction Output

What will be the output of the following Python code snippet using statsmodels ARIMA to fit and predict a simple time series?

ML Python

import numpy as np
from statsmodels.tsa.arima.model import ARIMA

data = np.array([1, 2, 3, 4, 5, 6, 7, 8, 9, 10])
model = ARIMA(data, order=(1,1,0))
model_fit = model.fit()
pred = model_fit.predict(start=9, end=10)
print(pred.tolist())

A [nan, nan]

B [9.0, 10.0]

C [10.0, 11.0]

D [8.0, 9.0]

💡 Hint

Remember that differencing reduces the series length by 1, and prediction starts at index 9.

The ARIMA(1,1,0) model fits the differenced data with one autoregressive term. Predicting at indices 9 and 10 gives the next values continuing the trend, which are 10.0 and 11.0.

❓ Hyperparameter

advanced

Choosing ARIMA Parameters

You have a time series with a strong seasonal pattern repeating every 12 months. Which ARIMA model order is most appropriate to capture this seasonality?

A (1, 1, 1)

B (2, 0, 2)

C (0, 1, 1)

D (1, 1, 1)(1, 0, 0, 12)

💡 Hint

Seasonal ARIMA models include extra parameters for seasonality.

Option D includes seasonal parameters (1,0,0) with a seasonal period of 12, which is needed to model yearly seasonality.

❓ Metrics

advanced

Evaluating ARIMA Model Performance

Which metric is most appropriate to evaluate the accuracy of an ARIMA model's forecast on a continuous time series?

A Accuracy score

B Mean Absolute Error (MAE)

C F1 Score

D Confusion Matrix

💡 Hint

Consider metrics for continuous numeric predictions.

MAE measures the average absolute difference between predicted and actual values, suitable for continuous forecasts like ARIMA outputs.

🔧 Debug

expert

Diagnosing ARIMA Model Fit Issues

After fitting an ARIMA(2,1,2) model to your data, you notice the residuals show a clear pattern and are not white noise. What is the most likely cause?

A The data was not differenced enough to achieve stationarity.

B The moving average order should be zero for better fit.

C The residuals always show patterns in ARIMA models.

D The model order is too low to capture the data's complexity.

💡 Hint

Check if the data is stationary before fitting ARIMA.

If residuals show patterns, it often means the model did not fully remove trends or seasonality, indicating insufficient differencing.