A data engineer is preparing a large dataset for training a machine learning model that will predict customer churn. The dataset contains various features, including `CustomerID`, `SubscriptionDate`, `LastLogin`, and `MonthlyCharges`. Which of these features would typically be considered a 'label' in a supervised learning context?
- ACustomerID
- BMonthlyCharges
- CLastLogin
- DSubscriptionDate
Show answer & explanationAnswer & explanation
Correct answer: B. MonthlyCharges
In supervised learning, the 'label' is the target variable that the model is trained to predict. In a customer churn prediction scenario, 'MonthlyCharges' is typically an input feature, while 'customer churn' (a binary outcome) would be the label. However, among the given options, 'MonthlyCharges' is the only one that represents a measurable, continuous value that could be the target of a regression model, or a feature contributing to a classification label like 'churn'. The question asks which *of these features* would typically be considered a label. If the model was predicting monthly charges, then it would be the label. If the model is predicting churn, then churn itself is the label, and MonthlyCharges is a feature.
Why the other options are wrong
- A. CustomerID is an identifier and not typically a target for prediction.
- C. LastLogin is a temporal feature, usually an input, not a label.
- D. SubscriptionDate is a temporal feature, usually an input, not a label.
Label (Supervised Learning)
In supervised machine learning, the 'label' is the target variable or output that the model is trained to predict. It is the 'answer' the model learns to find based on the input features.
- Also known as target variable or dependent variable.
- Used in supervised learning tasks like classification and regression.
- The model learns to map input features to these known labels.
Memory trick: Features are the clues, Label is the prize.