Machine Learning Fundamentals — DU B. Voc – Web Designing Previous Year Question Papers

Parts of a question must be answered together. Q1. (a) (b) (c) What is the assumption in the Naive Bayes classifier? What is the difference between Linear Regression and Logistic Regression? What do you understand by standardization? Explain briefly. (d) WhatisAUC-ROC curve? (e) What is `training set' and `test set' in a Machine Leaming Model? How much data will you allocate for your Training and Test sets ? (f) Define FPR and TPR. (g) Briefly explain Logistic Regression with a diagram. (h) What is Regularization? Explain the difference between LI Regularization (Lasso) and L2 Regularization (Ridge). Q2. (a) What is Bias and variance in a Machine Leaming Model? (b) What are the different strategies which can be followed for feature subset selection ? (c) Q3. (a) What Are the Different Types of Machine Leaming algorithms? Explain each with examples. Consider a dataset with actual values y and predicted values y from a regression model as given below. Assess the performance of the model using : a) Mean absolute error b) Mean squared error c) Root mean squared error d)R2 S.No Actual Predicted value (y) value(y) (b) (c) Given a dataset with three input features (xl,x2,x3) and target attribute y which can take either a value `yes' or `no'. A logistic regression model is trained on the given training set to estimate the parameters as Po =0.1, Pi = 0.35, , P2 = 0.11 and P3 = 0.2. compute the probability of y = "yes" given the instance as (2,4,7). Describe the k-fold cross-validation method a suitable choice. What are the ideal choices for the value of k ? can we say that Leave-one out cross-validation is a special case of k-fold cross-validation ? justify your answer. Q4 (a) The following dataset contains information about eight students, including their (5) CGPA, assessment score, project points, and result status. Use the K-Nearest Neighbor (k-NN) algorithm with k = 3 to classify the result (pass/fail) for a new test instance having the following attributes: CGPA = 8.0, Assessment = 62, and Project Points = 8. (b) What is the use of Normalization in Machine learning ? Explain its technique with an (5) A machine learning based imperfect email spam classifier flags emails as spam or not. (5) Out of 500 emails, 120 emails are actually spam. The model detects 150 emails as spam, out of which only 100 are truly spam. Also, the model is unable to detect 20 spam emails. Draw the confusion matrix and calculate the following performance metrics for the classifier: Q5. (a) Answer the following questions :- 1. List down the steps to perform K-Means clustering. 2. Given the following data points, cluster them into two groups (K=2) using the K-means clustering algorithm. Data Points are given as R1(1.0, 1.0), R2(1.5, Initial Centroids are given as C1 = (1.0, 1.0), C2 = (5.0, 7.0). (b) A dataset contains information about weather conditions and whether people play tennis. Calculate the entropy of the dataset and determine which attribute (Outlook or