Question 1
Which variant of the Naive Bayes Classifier will be most suitable to predict the Gender in the given dataset?
CategoricalNB
BernoulliNB
MultinomialNB
GaussianNB
Machine Learning Practice End Term: 10 May 2026 (January 2026 term)
The IIT Madras BS Machine Learning Practice (MLP) End Term paper sat on 10 May 2026, in the January 2026 term: 26 questions for 101 marks in 180 minutes. Every question is below with its answer. Take it as a timed mock test to be marked, or read it through first.
- 26
- 101
- 180 min
- 9
- 8
- 9
Show answer
Correct answer
GaussianNB
Question 2
Consider the following time series analysis code:
In the above code, what is the primary role of the parameter d=1 in order=(1,1,1)?
It controls the number of lag observations in the model
It represents the number of times the data is differenced
It defines the size of the moving average window
It determines the forecast horizon
Show answer
Correct answer
It represents the number of times the data is differenced
Question 3
In the following question, an Assertion (A) is followed by a Reason (R). Choose the correct option from the choices given below: Assertion (A): In classification problems, precision should always be maximized over recall Reason (R): Recall is independent of false negatives
Both A and R are true, and R is the correct explanation of A
Both A and R are true, but R is not the correct explanation of A
A is true, but R is false
A is false, but R is true
Both A and R are false
Show answer
Correct answer
Both A and R are false
Question 4
Consider the following code snippet:
Which employees are included in filtered_df?
Sara, Mike, Anna, Tom
Sara, Mike, Anna
Mike, Anna, Tom
Sara, Anna, Tom
Show answer
Correct answer
Sara, Mike, Anna, Tom
Question 5
What is the output of the following code? (Enter upto 2 decimals)
Show answer
A written answer, not marked automatically.
Question 6
Consider the following adaboost models:
Which of the following statements are MOST accurate?
adb1 is less likely to overfit than adb2 because its base trees are shallower.
Increasing n_estimators always improves test accuracy for AdaBoost.
Learning rate affects the contribution of each base learner; a smaller learning rate always reduces training accuracy.
adb2 with deeper trees will have lower bias but higher variance than adb1.
Show answer
Correct answers
adb1 is less likely to overfit than adb2 because its base trees are shallower.
adb2 with deeper trees will have lower bias but higher variance than adb1.
Question 7
Consider the following code snippet:
Which of the following statements are true?
If the model reaches convergence in under 20 iterations, it is guaranteed that every data point is mapped to its nearest centroid.
The value for number of clusters is a hyperparameter that can be validated using a silhouette score or the elbow method.
The algorithm will execute a total of 20 independent runs using different starting centroid seed
The initialization process will select 9 data points entirely at random to serve as the starting centroids.
Show answer
Correct answers
If the model reaches convergence in under 20 iterations, it is guaranteed that every data point is mapped to its nearest centroid.
The value for number of clusters is a hyperparameter that can be validated using a silhouette score or the elbow method.
Question 8
Consider the following Python code snippet used for basic image processing and answer the question that follows:
Which of the following statements are False?
represents an image with 3 color channels
The maximum possible pixel intensity value in is 256
The resize operation applied to reduces the number of channels
is obtained by flipping from Left to Right
Show answer
Correct answers
The maximum possible pixel intensity value in is 256
The resize operation applied to reduces the number of channels
is obtained by flipping from Left to Right
Question 9
Identify which of the following lines (marked in comments) contain errors in the given code snippet:
Line1
Line2
Line3
Line4
Show answer
Correct answers
Line1
Line3
Line4
Question 10
Consider the following code snippet:
What is the predicted class for the test datapoint?
Show answer
A written answer, not marked automatically.
Question 11
Consider the following code snippet and answer the question that follows:
Enter the value obtained as output
Show answer
A written answer, not marked automatically.
Question 12
Consider the following lists containing the true values and the predicted values.
Calculate the f1_score for class 1. (Answer up to 2 decimals)
Show answer
A written answer, not marked automatically.
Question 13
Consider the following code snippet for a multi-class classification problem with three classes: 0, 1, and 2. The ground-truth labels are stored in , and the predicted labels are stored in .
What is the precision score for class 1?
Show answer
A written answer, not marked automatically.
Question 14
Which of the following transformations are commonly used as Data Augmentation techniques?
Random cropping of images during training
Normalizing pixel values to the range of [0,1]
Resizing all images to a fixed resolution
Rotation of images with a small angle range
Show answer
Correct answers
Random cropping of images during training
Rotation of images with a small angle range
Question 15
Analyze the following code snippet:
Which of the following statements are correct?
mlp2 has more trainable parameters than mlp1.
Changing random_state may change predictions even with the same dataset.
Using 'sgd' solver always gives higher accuracy than 'adam'.
'tanh' activation guarantees better performance than 'relu'.
Show answer
Correct answers
mlp2 has more trainable parameters than mlp1.
Changing random_state may change predictions even with the same dataset.
Question 16
Which of the following statements regarding Tfidf Vectorizer is true?
Setting ensures that a term appearing in all documents will have an IDF value equal to zero
Setting creates an additional copy of each token in lowercase in the vocabulary
Setting ensures that only the 1000 most frequent terms are kept in the vocabulary
Setting will lead to terms with document frequency strictly lesser than 3 being ignored while building vocabulary
Show answer
Correct answers
Setting ensures that only the 1000 most frequent terms are kept in the vocabulary
Setting will lead to terms with document frequency strictly lesser than 3 being ignored while building vocabulary
Question 17
Consider the following code snippet:
Identify the correct code snippet/s to retrive only employees from departments where the average salary > 65000.
Show answer
Correct answers
Question 18
Select all the false statements amongst the given statements:
RandomForest uses sampling without replacement for bagging.
Boosting reduces bias by sequentially training models
Bagging reduces bias by training multiple models on different subsets of data and averaging their predictions.
Boosting trains all base models independently in parallel.
Show answer
Correct answers
RandomForest uses sampling without replacement for bagging.
Bagging reduces bias by training multiple models on different subsets of data and averaging their predictions.
Boosting trains all base models independently in parallel.
Question 19
Which of the following implementations of LogisticRgression cannot be used to tackle multinomial multiclass problems?
Show answer
Correct answer
Question 20
What is the range of the metric that is being calculated by the following code snippet?
0 to 1
—
-1 to 1
—
Show answer
Correct answer
—
Question 21
Consider the following cross validation strategy:
What does the above mentioned cross val strategy do?
LOOCV will train the model 5 times.
LOOCV will train the model 1 time.
The code is wrong
Show answer
Correct answer
LOOCV will train the model 5 times.
Question 22
Consider the given 2 different lasso regression models trained on the same dataset:
Statements: 1. The coefficients of lasso2 are more likely to be smaller in magnitude than those of lasso1. 2. lasso1 may have fewer zero coefficients than lasso2. 3. The predictions of lasso1 on the training data will exactly match y. 4. Increasing alpha always increases model accuracy on unseen data. Which of the given statements are correct?
2 and 3
1 and 2
3 and 4
1 and 4
Show answer
Correct answer
1 and 2
Question 23
Consider the following dataset X with 6 samples and 3 features and answer the given subquestions.
What is the sum of all values of the first column of ? (Round off to two decimal places)
Show answer
A written answer, not marked automatically.
Question 24
Consider the following dataset X with 6 samples and 3 features and answer the given subquestions.
What is the output obtained while running ?
Show answer
A written answer, not marked automatically.
Question 25
We create the following ml pipeline:
Which of the following statements is true?
RandomizedSearchCV tries all possible combinations of hyperparameters in param_dist.
Only 4 random combinations of hyperparameters will be tested.
Changing random_state has no effect on the hyperparameter search results.
Changing cv to 8 will be computationally cheaper.
Show answer
Correct answer
Only 4 random combinations of hyperparameters will be tested.
Question 26
Consider the following dataset X with 6 samples and 3 features and answer the given subquestions.
Show answer
A written answer, not marked automatically.