MLS-C01 Question 77
Select 2A data science team is building a machine learning model using a dataset containing numerical features with varying ranges. Some features have values between 0 and 1, while others range from thousands to millions. The team notices the model's performance is suboptimal. Which preprocessing steps can help address this issue?
- A
Apply Min-Max scaling to normalize feature values between a specified range, such as 0 to 1.
- B
Standardize the dataset by subtracting the mean and dividing by the standard deviation for each feature.
- C
Use data augmentation techniques like flipping or rotating images to increase dataset size.
- D
Perform one-hot encoding on the numerical features to represent them as categorical variables.
- E
Remove features with higher numerical ranges from the dataset to ensure consistency.
Show answer and explanation
Correct answers: A, B
Explanation
When numerical features have varying ranges, models can struggle to converge or give undue importance to features with larger scales. Scaling techniques like Min-Max scaling and standardization normalize the feature values, improving model performance and stability. These preprocessing steps are essential for numerical data handling in machine learning pipelines.
- A. Correct.
Min-Max scaling ensures all numerical features are scaled to the same range, such as 0 to 1, which helps prevent features with larger ranges from dominating the model's training process.
- B. Correct.
Standardization (z-score normalization) transforms data to have a mean of 0 and a standard deviation of 1, ensuring features with different scales are treated equally by the model.
- C. Incorrect.
Data augmentation is typically used for image, text, or time-series data to increase dataset size and diversity. It does not address the issue of varying feature ranges in numerical data.
- D. Incorrect.
One-hot encoding is used to convert categorical variables into binary vectors. It is not relevant for dealing with numerical features of varying scales.
- E. Incorrect.
Removing features with large numerical ranges is not a recommended solution, as it can lead to loss of important information and reduced model performance.