Microsoft Azure AI Fundamentals (AI-900)Describe fundamental principles of machine learning on AzureMedium
A data scientist is preparing a dataset for a regression model that predicts house prices. One of the features, 'Area in SqFt', ranges from 500 to 10,000, while another feature, 'Number of Bedrooms', ranges from 1 to 8. The model is sensitive to the scale of input features. To ensure that both features contribute equally to the model's learning process and prevent features with larger values from dominating, which data preprocessing technique should be applied?
- AFeature Scaling
- BOne-Hot Encoding
- COutlier Removal
- DPrincipal Component Analysis (PCA)
Show answer & explanationAnswer & explanation
Correct answer: A. Feature Scaling
The scenario describes features with different scales ('Area in SqFt' vs. 'Number of Bedrooms') and a model sensitive to scale. Feature scaling (e.g., normalization or standardization) is precisely designed to bring all features to a similar range or distribution, ensuring they contribute equally and preventing larger-valued features from dominating.
Why the other options are wrong
- B. One-Hot Encoding is for converting categorical data, not numerical feature scales.
- C. Outlier Removal deals with extreme data points, not the inherent scale differences between features.
- D. Principal Component Analysis (PCA) is a dimensionality reduction technique, not primarily for scaling features.
Feature Scaling
Feature scaling is a data preprocessing technique used to standardize or normalize the range of independent variables or features of data. It ensures that all features contribute equally to the model's learning process.
- Prevents features with larger magnitudes from dominating the learning process.
- Essential for algorithms sensitive to feature scales (e.g., SVM, K-Nearest Neighbors, neural networks).
- Common methods include Min-Max Scaling (Normalization) and Standardization (Z-score normalization).
Memory trick: Scaling: Make sure all features are on the 'same scale' so none unfairly 'weigh' more.