Week 13 practice quiz

Why is exploratory data analysis (EDA) considered a critical step in data preprocessing?

• It reduces the need for advanced machine learning models such as deep learning and neural networks.

• It ensures data are collected under consistent laboratory conditions.

• It helps analysts understand data distributions, detect patterns, and identify anomalies early.

• It transforms non-tabular data directly into prediction scores.

What is a common risk of handling missing data by simply removing incomplete records from the dataset?

• It can reduce sample representativeness and introduce bias.

• It may lead to loss of potentially valuable information from partially complete cases.

• It increases the likelihood of overfitting complex models.

• It results in longer computational time during model training.

What concern is raised about relying on AI for qualitative data analysis in user studies?

• AI requires prohibitively large datasets to function properly.

• AI might hallucinate insights, undermining trust.

• AI cannot be used unless it is built using symbolic logic.

• AI reduces the efficiency of coding and annotation processes.

What is a major limitation identified in applying combined qualitative and quantitative methods to eye-tracking studies?

• The eye-tracking data are too imprecise to be useful in most scientific context.

• The combination of methods is discouraged in controlled lab settings due to computational restraints and data overload.

• Manual annotation and coding of eye-tracking data require substantial time and effort, making them difficult to scale.

• Current visualization tools do not automatically resolve all ambiguities in gaze data interpretation.

Which of the following best describes the main goal of predictive analytics?

• To generate descriptive statistics from historical and archival data.

• To explore data without a predefined hypothesis.

• To forecast future outcomes based on historical patterns.

• To visualize data for better user interaction.

According to Tyagi et al. (2022), which field is not mentioned in the article as a major application area of predictive analytics?

• Healthcare

• Retail

• Banking

• Astrophysics

What is the main limitation of the 'unawareness' fairness notion in machine learning?

• It requires the use of sensitive attributes directly in decision-making.

• It can still lead to biased outcomes due to redundant encoding of sensitive attributes in other features.

• It cannot guarantee equal outcomes across all demographic groups.

• It primarily relies on establishing fairness by using counterfactual causal models which are known to be unreliable.

What is a key challenge in identifying algorithmic bias using regression analysis?

• It may misinterpret bias due to omitted relevant variables.

• Sensitive attributes are always easy to change in datasets.

• Access to the source code of algorithms is always required.

• Regression analysis can never detect bias reliably.

Which splitting criterion is primarily used by the CART algorithm for classification tasks?

• Information Gain Ratio

• Gini Index

• Chi-Square Statistic

• Variance Reduction

What is the main purpose of post-pruning in decision tree algorithms?

• To increase the depth of the tree for better accuracy

• To stop the tree from growing beyond a fixed depth

• To reduce overfitting and improve generalization

• To convert continuous variables into categorical variables