IBM Data Science Practice Test

Which model is most suitable for a binary outcome prediction?

1. Convolutional neural networks

2. Linear regression

3. Logistic regression

4. Decision trees

The correct answer is: Logistic regression

The most suitable model for binary outcome prediction is logistic regression. This statistical method is specifically designed to handle situations where the dependent variable is binary, meaning it can take only two possible outcomes, such as "yes/no" or "success/failure." Logistic regression uses the logistic function to model the probability that a given input point belongs to a certain class, producing probabilities that range from 0 to 1. If the predicted probability exceeds a threshold (commonly set at 0.5), the model classifies the outcome as one class; otherwise, it classifies it as the other. This unique capability to directly estimate the likelihood of a binary outcome, alongside providing interpretable coefficients related to the input features, makes it a widely used and effective choice for this type of problem. While convolutional neural networks are powerful for image processing tasks, they are more complex and often unnecessary for simple binary outcome predictions. Linear regression is typically used for continuous outcomes and does not handle the binary nature of the dependent variable appropriately, as it can produce values outside the [0, 1] range. Decision trees can certainly be used for binary classification tasks, but they do not provide the probabilistic output and interpretability that logistic regression offers, making logistic regression