Feature Selection and Data Science

What does EDA stand for in data science?

• Exploratory Data Analysis

• Exploratory Data Analytics

• Effective Data Assessment

• Extended Data Analysis

Explain the concept of feature engineering in data science.

• Feature engineering has no impact on model performance

• Feature engineering is not necessary in data science

• Feature engineering is crucial in data science as it directly impacts the model's ability to learn and make accurate predictions.

• Feature engineering only involves data cleaning

What is the filter method in feature selection?

• Choosing features randomly

• Selecting features based on statistical properties like correlation or variance.

• Selecting features based on alphabetical order

• Picking features based on the length of their names

Describe the wrapper method in feature selection.

• The wrapper method in feature selection is not based on any machine learning algorithm

• The wrapper method in feature selection uses only domain knowledge to select features

• The wrapper method in feature selection randomly selects features

• The wrapper method in feature selection selects features based on the performance of a specific machine learning algorithm.

How does the embedded method work in feature selection?

• The embedded method selects features after model training is completed.

• The embedded method selects features during model training based on their importance.

• The embedded method selects features based on their alphabetical order.

• The embedded method selects features randomly without any criteria.

What is under sampling and how is it used in data science?

• Under sampling is used to increase class imbalance in the dataset.

• Under sampling is a technique to remove outliers from the dataset.

• Under sampling involves selecting all instances from the majority class to create a balanced dataset.

• Under sampling involves randomly selecting a subset of the majority class instances to create a more balanced dataset for model training.

Explain the concept of over sampling in data science.

• Balancing class distribution by generating synthetic samples from the minority class to match the majority class.

• Creating duplicate samples from the majority class to match the minority class.

• Removing samples from the majority class to match the minority class.

• Ignoring class distribution and using the original dataset as is.

What is an imbalance dataset and why is it a challenge in data science?

• The challenge of imbalanced datasets in data science is overhyped and not a real issue.

• Imbalanced datasets are easier to work with in data science due to their predictable nature.

• An imbalanced dataset is a dataset where the distribution of classes is uniform, leading to better model performance.

• An imbalanced dataset is a dataset where the distribution of classes is not uniform, making some classes have significantly more samples than others. This poses a challenge in data science because models trained on imbalanced datasets may have biases towards the majority class, leading to poor performance on minority classes.

How does ensemble method help in feature selection?

• Ensemble methods randomly select features without considering importance

• Ensemble methods ignore feature relevance in the selection process

• Ensemble methods only focus on one model for feature selection

• Ensemble methods consider feature importance across multiple models or trees to aid in feature selection.

Why is EDA important before performing feature selection in data science?

• To make the process more complicated

• To waste time

• To confuse the data scientists

• To understand data distribution, relationships between variables, identify outliers, and determine relevant features.