Databricks Machine Learning Associate Question 144
Select 4You are working on a dataset in Databricks to predict customer churn for a subscription service. During Exploratory Data Analysis (EDA), which of the following steps would help you identify issues with data quality that could impact your machine learning model?
- A
Checking for missing values in the dataset
- B
Visualizing the distribution of numerical features
- C
Splitting the dataset into training, validation, and test sets
- D
Analyzing correlations between features
- E
Inspecting for outliers in numerical columns
Show answer and explanation
Correct answers: A, B, D, E
Explanation
Exploratory Data Analysis involves understanding the dataset to uncover data quality issues that could impact the machine learning workflow. Steps like checking for missing values, visualizing distributions, analyzing correlations, and inspecting for outliers are crucial for identifying potential problems and ensuring the dataset is ready for modeling. Splitting the dataset is a separate step in the ML workflow, not a part of EDA.
- A. Correct.
Missing values can lead to biased or incorrect model results, so identifying them is critical during EDA.
- B. Correct.
Visualizing the distribution of numerical features can highlight potential issues like skewness or irregular distributions that may need to be addressed.
- C. Incorrect.
Splitting the dataset is a step in the ML workflow, but it is not directly part of identifying data quality issues during EDA.
- D. Correct.
Analyzing correlations between features helps in identifying redundant or highly correlated features that could affect model performance.
- E. Correct.
Inspecting for outliers is important for detecting extreme values that could distort the model training process.