Data distributions and outliers in ML Python - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

Understanding the effect of outliers on mean and median

Consider a dataset with values: [10, 12, 14, 15, 100]. Which statement best describes the effect of the outlier (100) on the mean and median?

A The mean is heavily influenced by the outlier, but the median remains relatively stable.

B Both mean and median are equally influenced by the outlier.

C The median is heavily influenced by the outlier, but the mean remains stable.

D Neither mean nor median is influenced by the outlier.

❓ Predict Output

intermediate

Detecting outliers using the IQR method

What is the output of the following Python code that detects outliers using the interquartile range (IQR)?

ML Python

import numpy as np

data = np.array([5, 7, 8, 12, 15, 18, 22, 100])
Q1 = np.percentile(data, 25)
Q3 = np.percentile(data, 75)
IQR = Q3 - Q1
outliers = data[(data < Q1 - 1.5 * IQR) | (data > Q3 + 1.5 * IQR)]
print(outliers.tolist())

A []

B [5, 7]

C [100]

D [22, 100]

❓ Model Choice

advanced

Choosing a model robust to outliers

You have a dataset with many outliers. Which regression model is best suited to minimize the effect of outliers on predictions?

A Robust regression using Huber loss

B Ridge regression with L2 regularization

C Linear regression using least squares loss

D Polynomial regression of degree 5

❓ Hyperparameter

advanced

Adjusting hyperparameters to handle outliers in clustering

When using DBSCAN clustering on data with noise and outliers, which hyperparameter adjustment helps to better identify outliers as noise points?

A Decrease the epsilon (eps) value to make clusters tighter

B Decrease the minimum samples (min_samples) to form a cluster

C Increase the epsilon (eps) value to include more points in clusters

D Increase the minimum samples (min_samples) to require more points per cluster

❓ Metrics

expert

Evaluating model performance with imbalanced data and outliers

You have a classification dataset with imbalanced classes and some outliers. Which metric is most reliable to evaluate model performance in this scenario?

A Accuracy

B F1-score

C Precision

D Mean Squared Error (MSE)