MLA-C01 Question 50
Select 3You are building a machine learning model to predict housing prices using a dataset that contains information about homes, such as square footage, number of bedrooms, and location. During your exploratory data analysis, you notice the following issues: 1) The 'square footage' feature has some missing values, 2) The 'price' feature contains extreme outliers, and 3) There are duplicate rows in the dataset. What are the most appropriate data cleaning and transformation steps to address these issues?
- A
Impute missing values in the 'square footage' feature using the mean of the available values.
- B
Remove extreme outliers in the 'price' feature by filtering out values beyond three standard deviations from the mean.
- C
Remove duplicate rows from the dataset to ensure data integrity.
- D
Replace missing values in the 'square footage' feature with zero to simplify preprocessing.
- E
Normalize the 'price' feature to bring its values within a common scale.
Show answer and explanation
Correct answers: A, B, C
Explanation
To prepare the dataset for machine learning, it is essential to address missing values, outliers, and duplicates. Imputing missing values with the mean ensures continuity in the dataset. Removing outliers prevents them from skewing the model's predictions, and eliminating duplicate rows ensures data integrity. Replacing missing values with zero is not appropriate here, as it can mislead the model, and normalization does not address the specific issues described in the scenario.
- A. Correct.
Correct: Imputing missing values with the mean is a common data cleaning technique that preserves most of the data without introducing bias.
- B. Correct.
Correct: Removing extreme outliers is an effective way to ensure the model is not overly influenced by anomalous data points.
- C. Correct.
Correct: Duplicate rows can distort analyses and predictions, so removing them is a standard data cleaning step.
- D. Incorrect.
Incorrect: Replacing missing values with zero can introduce a bias and misrepresent the data, especially for numerical features like square footage.
- E. Incorrect.
Incorrect: Normalization is a feature scaling technique, not a method to handle outliers, missing values, or duplicates.