Home / Past papers / MLP End Term / this paper

IIT M FOUNDATION DIPLOMA FN EXAM - 01 Sep 2024

MLP End Term37 questions100 marks20 MCQ · 11 SA · 6 MSQFoundation / Diploma / forenoon

Elapsed 00:00

Comprehension passage - the questions below it refer to this

Consider following common data and answer the subquestions: question figure

Consider following common data and answer the subquestions:

Q1SA2 marks

What will be the output of the following code snippet? question figure

What will be the output of the following code snippet? Enter -1, if you think the above code snippet will generate an error.

Your numeric answer:

Q2SA2 marks

What will be the output of the following code snippet? question figure

What will be the output of the following code snippet? Enter -1, if you think the above code snippet will generate an error.

Your numeric answer:

Q3SA2 marks

What will be the output of the following code snippet? question figure

What will be the output of the following code snippet? Enter -1, if you think the above code snippet will generate an error.

Your numeric answer:

Q4SA2 marks

How many columns have datatype as ”float64”?

Your numeric answer:

Q5SA2 marks

What will be the output of the following code snippet? question figure

What will be the output of the following code snippet? Enter -1, if you think the above code snippet will generate an error.

Your numeric answer:

Q6SA2 marks

Consider following parameter grid for regression with RandomForest: question figure

Consider following parameter grid for regression with RandomForest: How many unique combinations of hyperparameters are there in the above parameter grid?

Your numeric answer:

Q7MCQ1 mark

How to make SGDRegressor stop after 1000 epochs?

Q8MCQ2 marks

DictVectorizer
HashingVectorizer
FeatureHasher

Q9MCQ2 marks

Which of the following is an example of an ordinal variable?

Blood type (A, B, AB, O)
Satisfaction rating (Very Unsatisfied, Unsatisfied, Neutral, Satisfied, VerySatisfied)
Height of individuals
Gender

Q10MCQ2 marks

Q11MCQ2 marks

cv
reg_rate
alpha
tol

Q12MCQ3 marks

1000
100
99
999

Q13MCQ3 marks

Q14MCQ3 marks

It determines whether to include a bias term in the model.
It sets the degree of the polynomial features.
It controls the regularization strength.
It specifies the number of features to consider when fitting the model.

Q15MCQ3 marks

To specify the maximum number of samples per leaf.
To limit the depth of the tree.
To set the minimum number of samples required to split an internal node.
To control the minimum impurity decrease required for a split.

Q16MCQ3 marks

It determines the function to measure the quality of a split.
It specifies the number of features to consider when looking for the best split.
It sets the strategy used to choose the split at each node.
It controls whether to use the Gini impurity or entropy.

Q17MCQ3 marks

Q18MCQ3 marks

Consider below statements and choose the correct option: Statement1:The bagging technique combines multiple models trained on different subsets of data Statement2:The Boosting technique trains the model Parallelly, focusing on the error made by the other model.

Statement1 is True, Statement2 is True
Statement1 is True, Statement2 is False
Statement1 is False, Statement2 is True
Statement1 is False, Statement2 is False

Q19MCQ3 marks

How does agglomerative clustering handle outliers?

It ignores outliers during the clustering process.
It assigns outliers to the nearest cluster.
It creates separate clusters for outliers.
It removes outliers from the dataset before clustering.

Q20MCQ3 marks

What is agglomerative clustering?

A hierarchical clustering technique that starts with each data point as itscluster and merges the closest clusters iteratively.
A method for partitioning data into a predefined number of clusters.
A clustering algorithm that uses centroids to iteratively assign data points toclusters.
A dimensionality reduction technique that projects data onto a lower-dimensional space.

Q21MCQ3 marks

The distance between cluster centroids
The number of clusters formed
The within-cluster sum of squared distances
The silhouette coefficient

Q22MCQ4 marks

Consider the following ML task/steps for a regression dataset: 1. Read the data from a file (named ‘dataset.csv’). It has 7 columns. The last column is the target variable, all 6 features numerical. 2. Remove rows which has target values missing. 3. Split the data into training and test sets. Take randomly the 70% of rows in the training set and the rest of them into the test set. 4. Fill the missing values in the features by KNNImputer using 3 nearest neighbours. 5. Train a LinearRegression model on the training set. 6. Report R2 score on the test set. Which of the following code snippets correctly accomplishes the above task? Assume necessary imports.

Q23MCQ4 marks

Q24MCQ4 marks

Q25MCQ4 marks

The predictions are based on the majority vote.
The predictions are based on the average of probabilities predicted by eachclassifier.
The predictions are based on the weighted sum of the predictions.
The predictions are based on the classifier with the highest accuracy

Q26MCQ4 marks

5 centroids are randomly initialized 10 times
10 centroids are randomly initialized 5 times
5 samples in the dataset are selected as initialization point such that they areat least 10 units away from each other
10 samples in the dataset are selected as initialization point such that they areat least 5 units away from each other

Q27MSQ2 marksselect all that apply

Data snooping:

Leads to biased estimation on test sets
Increases the risk of false positives
Leads to better estimation on training sets
Reduces the risk of false positives

Q28MSQ2 marksselect all that apply

Which of the following algorithms may get impacted by feature scaling?

LinearRegression
DecisionTree
SVM
NaiveBayes

Q29MSQ3 marksselect all that apply

The logistic regression model will use equal weights for both the classes in animbalanced dataset.
The model does not use any regularization because the parameter C is set.
The “balanced” mode uses the values of y to automatically adjust weightsinversely proportional to class frequencies in the input data.
The value of C indicates that the model will apply a regularization.

Q30MSQ3 marksselect all that apply

Which of the following statements are true?

KNeighborsClassifier with high values of n_neighbors produces complexdecision boundaries.
KNeighborsClassifier with high values of n_neighbors produces smoothdecision boundaries.
In KNeighborsClassifier the scale of the features(columns) can impact thedecision boundaries.
None of these

Q31MSQ3 marksselect all that apply

Which options are correct for Support Vectors in SVM ?

Support vectors are the data points nearest to the hyperplane
Using these support vectors, we maximize the margin of the classifier.
Using these support vectors, we minimize the margin of the classifier.
None of these

Q32MSQ3 marksselect all that apply

Which of the following are correct for Kmeans clustering algorithm?

It only finds spherical clusters.
The output of the algorithms depends upon initial cluster centroids.
It can automatically detect most appropriate value of k.
None of these.

Q33SA3 marks

Your numeric answer:

Q34SA3 marks

Your numeric answer:

Q35SA3 marks

Your numeric answer:

Comprehension passage - the questions below it refer to this

Based on the above data, answer the given subquestions. question figure

Q36SA2 marks

What will be the output of the following code? question figure

What will be the output of the following code?

Your numeric answer:

Q37SA2 marks

What will be the output of the following code? question figure

What will be the output of the following code?

Your numeric answer:

Answer key

Q1 (SA, 2 marks): 24
Q2 (SA, 2 marks): 3
Q3 (SA, 2 marks): 34
Q4 (SA, 2 marks): 2
Q5 (SA, 2 marks): 64
Q6 (SA, 2 marks): 21
Q7 (MCQ, 1 marks): C
Q8 (MCQ, 2 marks): A
Q9 (MCQ, 2 marks): B
Q10 (MCQ, 2 marks): A
Q11 (MCQ, 2 marks): C
Q12 (MCQ, 3 marks): A
Q13 (MCQ, 3 marks): B
Q14 (MCQ, 3 marks): A
Q15 (MCQ, 3 marks): B
Q16 (MCQ, 3 marks): A
Q17 (MCQ, 3 marks): D
Q18 (MCQ, 3 marks): B
Q19 (MCQ, 3 marks): B
Q20 (MCQ, 3 marks): A
Q21 (MCQ, 3 marks): C
Q22 (MCQ, 4 marks): A
Q23 (MCQ, 4 marks): B
Q24 (MCQ, 4 marks): E
Q25 (MCQ, 4 marks): B
Q26 (MCQ, 4 marks): A
Q27 (MSQ, 2 marks): A, B
Q28 (MSQ, 2 marks): A, C
Q29 (MSQ, 3 marks): C, D
Q30 (MSQ, 3 marks): B, C
Q31 (MSQ, 3 marks): A, B
Q32 (MSQ, 3 marks): A, B
Q33 (SA, 3 marks): 0.25
Q34 (SA, 3 marks): 1
Q35 (SA, 3 marks): 3
Q36 (SA, 2 marks): 1
Q37 (SA, 2 marks): 0.64 to 0.66

Concepts this paper tests

Auto-generated summary from the source site. Handy as a checklist; the lessons above explain each idea from scratch.

Here’s a structured summary of core topics, concepts, principles, and equations essential for the Diploma Level: Machine Learning Practice (Computer-Based Exam). The exam covers Pandas, Scikit-Learn, ML algorithms, hyperparameter tuning, evaluation metrics, and clustering. Key areas are grouped into theoretical concepts, practical coding skills, and mathematical foundations (where applicable).

1. Core Topics & Concepts

A. Data Handling with Pandas

DataFrame Operations:
- df.shape, df.isna().sum(), df.dropna(), df.fillna().
- df.iloc[] (integer-location based indexing), df.loc[] (label-based indexing).
- df.value_counts() for categorical data.
- Handling missing values (np.nan).
Key Equations:
- Non-missing values in DataFrame:

\text{Non-missing values} = (\text{rows} \times \text{columns}) - \text{sum of missing values}

Mean of a column (ignoring NaN):

\text{Mean} = \frac{\sum_{i=1}^{n} x_i}{\text{count of non-NaN values}}

B. Machine Learning Algorithms

1. Supervised Learning

Linear Models:
- Linear Regression: Predicts continuous output. Uses OLS (Ordinary Least Squares).
- Ridge Regression (L2 Regularization):

\text{Loss} = \text{MSE} + \alpha \sum_{i=1}^{n} w_i^2

alpha=0: No regularization.
- Lasso Regression (L1 Regularization):

\text{Loss} = \text{MSE} + \alpha \sum_{i=1}^{n} |w_i|

SGDRegressor: Stochastic Gradient Descent for linear models.
- max_iter: Number of epochs (default: 1000).
- Logistic Regression:
Predicts probabilities for classification.
class_weight='balanced': Adjusts weights inversely proportional to class frequencies.
C: Inverse of regularization strength (smaller C = stronger regularization).
Predict probabilities: predict_proba(X).
- Support Vector Machines (SVM):
Support Vectors: Data points closest to the hyperplane.
Margin Maximization: Maximize distance between hyperplane and support vectors.
Kernel Trick: For non-linear decision boundaries.
- Decision Trees:
Parameters:
- max_depth: Limits tree depth to prevent overfitting.
- min_samples_split: Minimum samples required to split a node.
- min_samples_leaf: Minimum samples required at a leaf node.
- criterion: 'gini' (Gini impurity) or 'entropy' (information gain).
Overfitting Risk: Deeper trees capture noise; shallower trees generalize better.
- Random Forest:
Ensemble of Decision Trees.
Bootstrapping: Samples with replacement (bootstrap=True).
Feature Subsampling: max_features controls randomness.
Out-of-Bag (OOB) Error: Error estimated on unseen samples.
- K-Nearest Neighbors (KNN):
Lazy Learner: No training; memorizes data.
Distance Metric: Euclidean, Manhattan, etc.
n_neighbors:
- High n_neighbors: Smoother decision boundaries.
- Low n_neighbors: Complex boundaries (risk of overfitting).
Feature Scaling: Critical (distance-based).
- Perceptron:
Linear classifier for binary classification.
Decision Rule:

\hat{y} = \text{sign}(w^T x + b)

2. Unsupervised Learning

K-Means Clustering:
- Objective: Minimize within-cluster sum of squares (inertia).
- Inertia:

\text{Inertia} = \sum_{i=1}^{n} \min_{\mu_j \in C} ||x_i - \mu_j||^2

Initialization: 'random' or 'k-means++'.
Limitations:
- Assumes spherical clusters.
- Sensitive to initial centroids.
- Agglomerative Clustering:
Hierarchical: Starts with each point as a cluster; merges closest clusters.
Handles Outliers: Assigns outliers to the nearest cluster.

3. Ensemble Methods

Bagging (Bootstrap Aggregating):
- Trains multiple models on bootstrapped samples.
- Reduces Variance (e.g., Random Forest).
Boosting:
- Sequentially trains models, focusing on errors from previous models.
- Examples: AdaBoost, Gradient Boosting.
Voting Classifier:
- Hard Voting: Majority class label.
- Soft Voting: Average of predicted probabilities.

C. Model Evaluation

Metrics:
- Classification:
  - Accuracy: $\frac{\text{TP} + \text{TN}}{\text{TP} + \text{TN} + \text{FP} + \text{FN}}$
  - Precision: $\frac{\text{TP}}{\text{TP} + \text{FP}}$
  - Recall (Sensitivity): $\frac{\text{TP}}{\text{TP} + \text{FN}}$
  - F1-Score: Harmonic mean of precision and recall.
  - Confusion Matrix: TP, TN, FP, FN.
- Regression:
  - R² Score: $1 - \frac{\text{SS}_{\text{res}}}{\text{SS}_{\text{tot}}}$
  - MSE (Mean Squared Error): $\frac{1}{n} \sum_{i=1}^{n} (y_i - \hat{y}_i)^2$
Cross-Validation:
- k-Fold CV: Splits data into k folds; trains on k-1 folds, tests on 1.
- Leave-One-Out CV (LOOCV): n models for n samples.
Dummy Classifier:
- Baseline model (e.g., strategy='most_frequent').

D. Preprocessing

Feature Scaling:
- StandardScaler: $\frac{x - \mu}{\sigma}$ (mean=0, std=1).
- MinMaxScaler: $\frac{x - \min(X)}{\max(X) - \min(X)}$ (range [0, 1]).
- Impacted Algorithms: KNN, SVM, Linear Regression, Neural Networks.
Handling Missing Data:
- Imputation: KNNImputer, SimpleImputer.
- Dropping: df.dropna().
Feature Extraction:
- DictVectorizer: Converts dictionaries to feature matrices.
- PolynomialFeatures: Generates polynomial/combination features.
  - include_bias: Adds intercept term.

E. Hyperparameter Tuning

Grid Search:
- Exhaustive search over specified parameter values.
- Example:
```
param_grid = {'n_estimators': [10, 20, 30], 'max_features': [2, 3, 4]}
```
- Total Combinations: Product of parameter options (e.g., 3 * 3 = 9).
Random Search: Randomly samples hyperparameters.

F. Bias-Variance Tradeoff

Bias: Error due to overly simplistic assumptions (underfitting).
Variance: Error due to excessive complexity (overfitting).
Regularization: Controls model complexity (e.g., alpha in Ridge/Lasso).

G. Data Leakage & Best Practices

Data Snooping: Peeking at test data leads to biased estimates.
Train-Test Split: train_test_split(test_size=0.3, random_state=42).
Pipelines: Ensure preprocessing is fitted only on training data.

2. Key Equations

Concept	Equation
Linear Regression	$\hat{y} = w^T x + b$
Ridge Regression	$\text{Loss} = \text{MSE} + \alpha \sum w_i^2$
Lasso Regression	$\text{Loss} = \text{MSE} + \alpha \sum \|w_i\|$
Logistic Regression	$P(y=1) = \frac{1}{1 + e^{-(w^T x + b)}}$
Precision	$\frac{\text{TP}}{\text{TP} + \text{FP}}$
Recall	$\frac{\text{TP}}{\text{TP} + \text{FN}}$
F1-Score	$2 \times \frac{\text{Precision} \times \text{Recall}}{\text{Precision} + \text{Recall}}$
R² Score	$1 - \frac{\text{SS}_{\text{res}}}{\text{SS}_{\text{tot}}}$
Inertia (K-Means)	$\sum_{i=1}^{n} \min_{\mu_j} \|\|x_i - \mu_j\|\|^2$
Gini Impurity	$1 - \sum_{i=1}^{C} p_i^2$
Entropy	$-\sum_{i=1}^{C} p_i \log(p_i)$
Euclidean Distance	$\sqrt{\sum_{i=1}^{n} (x_i - y_i)^2}$

Graph 2: Unsupervised Learning

Graph 3: Model Evaluation

Graph 4: Preprocessing & Feature Engineering

Graph 5: Ensemble Methods

Graph 6: Hyperparameter Tuning

4. Exam Tips

Pandas:
- Know how to count non-missing values (df.shape[0]*df.shape[1] - df.isna().sum().sum()).
- Use value_counts() for categorical data.
Scikit-Learn:
- Syntax: model.fit(X_train, y_train), model.predict(X_test).
- Pipelines: Ensure preprocessing is only fitted on training data.
Algorithms:
- Decision Trees: max_depth controls overfitting.
- Random Forest: n_estimators = number of trees.
- KNN: Scale features; n_neighbors affects boundaries.
Evaluation:
- Precision/Recall Tradeoff: High precision = fewer FP; high recall = fewer FN.
- Cross-Validation: LOOCV generates n models for n samples.
Clustering:
- K-Means: Inertia measures compactness.
- Agglomerative: Assigns outliers to nearest clusters.
Hyperparameters:
- Ridge/Lasso: alpha controls regularization strength.
- SGD: max_iter = epochs.