Home / Past papers / MLP End Term / this paper

IIT M IMPROVEMENT FN EXAM QDD1 22 Dec - 22 Dec 2024

MLP End Term37 questions100 marks21 MCQ · 12 MSQ · 4 SAImprovement / forenoon / QDD1

Elapsed 00:00

Comprehension passage - the questions below it refer to this

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

Q1MCQ1 mark

Which of the following commands will return the average age of employees?

Q2MSQ3 marksselect all that apply

Which of the following commands will return the number of employees in the 'HR ’ department?

Q3MSQ3 marksselect all that apply

Which of the following methods can be used to filter employees with a salary greater than 50,000?

Q4MCQ1 mark

What does the following code return? question figure

What does the following code return?

The row with the employee who has the lowest performance score.
The row with the employee who has the highest performance score.
The row with the highest salary.
None of these.

Q5MCQ4 marks

Which of the following code snippets will plot the learning curve for training over 100 epochs? Assume that all necessary imports, parameter settings, and initializations are correctly done, and the variable names suggest the data they hold or point to.

Q6MCQ4 marks

Q7MCQ3 marks

Consider the following code: question figure

Consider the following code:

Q8MCQ3 marks

Q9MCQ3 marks

Q10MCQ3 marks

Output score is likely to increase.
Output score is likely to decrease.
Output score may increase or decrease depending on other factors.
Output score will remain the same.

Q11MCQ3 marks

The number of base estimators to be trained.
The size of each sample batch for testing.
The maximum number of features used for training.
The number of samples to draw from X to train each base estimator

Q12MCQ3 marks

Performs hierarchical clustering and outputs cluster distances.
Outputs the silhouette score for the data points.
Groups data points into 2 clusters and outputs the labels of the clusters foreach data point.
Reduces the dimensionality of the dataset to 2 dimensions.

Q13MCQ3 marks

It increases the model’s capacity to fit the training data closely, potentiallycausing overfitting.
It adds a stronger regularization term to the loss function, reducing themodel’s complexity and helping prevent overfitting.
It makes the model more prone to overfitting by increasing its sensitivity tonoise in the training data.
It directly affects the learning rate, controlling how fast the model convergesduring training.

Q14MCQ2 marks

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

What will be the output of the following code?

Q15MCQ2 marks

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

What will be the output of the following code?

[5, 6]
[4, 5]
[3, 4]
[2, 3]

Q16MCQ2 marks

You have loaded a dataset with 500 samples and 50 features into a Pandas DataFrame. Some samples have missing values for a few features. If you want to remove rows with more than 5 missing values, what method would you use?

Q17MCQ2 marks

Statement 1 False, Statement 2 False
Statement 1 True, Statement 2 False
Statement 1 False, Statement 2 True
Both statements are True

Q18MCQ2 marks

Q19MCQ2 marks

What will be the output of the code? question figure

1.25
2.56
4.25
3.25
Given code will return an error.

Q20MCQ2 marks

It adjusts the learning rate based on the number of iterations.
It specifies the learning rate to decrease as the model converges.
It specifies the fraction of the dataset used for each step in the stochasticgradient descent.
It defines the initial value for the learning rate and keeps it constantthroughout training.

Q21MCQ2 marks

Which of the following is NOT a hyperparameter in machine learning?

Learning rate in gradient descent
Bias in a linear regression model
Number of clusters in k-means
Minimum leaf size in decision trees

Q22MCQ2 marks

Q23MCQ2 marks

It decreases the variance among individual trees.
It increases the variance among individual trees.
It reduces the number of samples used for training each tree.
It reduces overfitting by restricting tree depth.

Q24MSQ4 marksselect all that apply

Consider the following code: question figure

Consider the following code:

[47, 31, 18, 95, 77]
[31, 41, 85, 31]
[47, 31, 18]
[47, 31, 85, 77]
[31, 41, 18, 95, 85, 77]
[18, 95, 85, 77]

Q25MSQ4 marksselect all that apply

Q26MSQ4 marksselect all that apply

The neural network contains 3 hidden layers with 5 neurons in each hiddenlayer
The neural network contains 5 hidden layers with 3 neurons in each hiddenlayer
The neural network contains 2 hidden layers with 3 neurons in the secondhidden layer
The neural network contains 2 hidden layers with 5 neurons in the first hiddenlayer
None of the given options are correct

Q27MSQ3 marksselect all that apply

Q28MSQ3 marksselect all that apply

GaussianNB
LogisticRegression
SGDClassifier
RandomForestClassifier

Q29MSQ3 marksselect all that apply

Q30MSQ3 marksselect all that apply

What parameters of the k-means clustering algorithm should be tuned to optimize the model performance?

Q31MSQ3 marksselect all that apply

Consider the following options for evaluating the optimal number of clusters in k-means. Which approaches are valid?

Using an Elbow Curve.
Computing Silhouette Scores.
Cross-validation with multiple folds.
All of these

Q32MSQ3 marksselect all that apply

Which of the following statements are correct about the Silhouette Coefficient?

It measures how well each sample is clustered.
A higher score indicates better-defined clusters.
It requires ground truth labels (target variable) to compute.
It can be used to evaluate the quality of any clustering algorithm.

Q33MSQ2 marksselect all that apply

Q34SA3 marks

Your numeric answer:

Q35SA3 marks

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

What will be the output of the following code snippet?

Your numeric answer:

Q36SA3 marks

Your numeric answer:

Q37SA2 marks

Your numeric answer:

Answer key

Q1 (MCQ, 1 marks): D
Q2 (MSQ, 3 marks): A, B
Q3 (MSQ, 3 marks): A, B, C
Q4 (MCQ, 1 marks): B
Q5 (MCQ, 4 marks): A
Q6 (MCQ, 4 marks): A
Q7 (MCQ, 3 marks): A
Q8 (MCQ, 3 marks): B
Q9 (MCQ, 3 marks): A
Q10 (MCQ, 3 marks): A
Q11 (MCQ, 3 marks): D
Q12 (MCQ, 3 marks): C
Q13 (MCQ, 3 marks): B
Q14 (MCQ, 2 marks): D
Q15 (MCQ, 2 marks): A
Q16 (MCQ, 2 marks): C
Q17 (MCQ, 2 marks): D
Q18 (MCQ, 2 marks): C
Q19 (MCQ, 2 marks): B
Q20 (MCQ, 2 marks): D
Q21 (MCQ, 2 marks): B
Q22 (MCQ, 2 marks): C
Q23 (MCQ, 2 marks): B
Q24 (MSQ, 4 marks): A, B, C, E
Q25 (MSQ, 4 marks): A, C
Q26 (MSQ, 4 marks): C, D
Q27 (MSQ, 3 marks): A, C, D
Q28 (MSQ, 3 marks): A, C
Q29 (MSQ, 3 marks): A, B
Q30 (MSQ, 3 marks): A, B, C
Q31 (MSQ, 3 marks): A, B
Q32 (MSQ, 3 marks): A, B, D
Q33 (MSQ, 2 marks): A, B, D
Q34 (SA, 3 marks): 1
Q35 (SA, 3 marks): 1
Q36 (SA, 3 marks): 1
Q37 (SA, 2 marks): 1.0

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), along with Mermaid knowledge graphs to visualize relationships.

1. Core Topics & Concepts

A. Data Preprocessing

Key Concepts:

Handling Missing Data
- df.dropna(thresh=N): Drops rows with fewer than N non-NaN values.
- df.fillna(value): Fills missing values.
Feature Scaling
- StandardScaler: Standardizes features (mean=0, std=1).
  - Formula:

z = \frac{x - \mu}{\sigma}

MinMaxScaler: Scales features to a range (default: [0, 1]).
- Formula:

x_{\text{scaled}} = \frac{x - x_{\text{min}}}{x_{\text{max}} - x_{\text{min}}}

Encoding Categorical Data
- OneHotEncoder: Converts categorical features into binary columns.
  - Example: ['cat', 'dog'] → [[1, 0], [0, 1]].
- LabelEncoder: Converts categories to integers (e.g., ['cat', 'dog'] → [0, 1]).
Polynomial Features
- PolynomialFeatures(degree=N): Generates polynomial combinations up to degree N.
  - Example: X = [a, b] → [1, a, b, a², ab, b²] for degree=2.

Key Pandas/NumPy Commands:

Task	Command
Compute mean of a column	`df['column'].mean()`
Filter rows	`df[df['Salary'] > 50000]`
Count values in a column	`df['Department'].value_counts()['HR']`
Sort by a column	`df.sort_values(by='Score', ascending=False)`
Select top row after sorting	`df.sort_values(...).iloc[0]`

B. Supervised Learning

1. Linear Models

Linear Regression
- Equation:

y = \beta_0 + \beta_1 x_1 + \beta_2 x_2 + \dots + \beta_n x_n + \epsilon

SGDRegressor:
- learning_rate='constant': Keeps learning rate fixed (eta0).
- partial_fit(X, y): Incremental learning (useful for large datasets).
- Logistic Regression
Sigmoid function:

\sigma(z) = \frac{1}{1 + e^{-z}}

SGDClassifier: Supports partial_fit for online learning.

2. Decision Trees

Parameters:
- max_depth: Maximum depth of the tree.
- min_samples_split: Minimum samples required to split a node.
- criterion: 'gini' (Gini impurity) or 'entropy' (information gain).
Attributes:
- tree_.max_depth: Actual depth of the tree.
- feature_importances_: Importance of each feature.
Maximum Leaf Nodes in a Balanced Binary Tree:

\text{Leaf nodes} = 2^{\text{max\_depth}}

Example: max_depth=3 → $2^3 = 8$ leaf nodes.

3. Ensemble Methods

Random Forest
- max_features: Number of features to consider for splits (lower → higher variance).
- n_estimators: Number of trees in the forest.
Gradient Boosting (GradientBoostingClassifier)
- n_estimators: Number of boosting stages.
- learning_rate: Shrinks contribution of each tree.
Bagging (BaggingClassifier)
- max_samples: Number of samples to draw for each base estimator (default: 1.0 = all samples).

4. Neural Networks (`MLPClassifier`/`MLPRegressor`)

Parameters:
- hidden_layer_sizes: Tuple specifying neurons per layer (e.g., (5, 3) = 2 layers with 5 and 3 neurons).
- alpha: L2 regularization strength (higher → less overfitting).
- max_iter: Maximum epochs.
Effect of alpha:
- Higher alpha → stronger regularization → simpler model.

5. Model Evaluation

Metrics:
- Regression:
  - Mean Squared Error (MSE):

\text{MSE} = \frac{1}{n} \sum_{i=1}^n (y_i - \hat{y}_i)^2

R² Score:

R^2 = 1 - \frac{\text{SS}_{\text{res}}}{\text{SS}_{\text{tot}}}

Classification:
- Confusion Matrix: TP, TN, FP, FN.
- ROC-AUC: Area under the ROC curve.
- Log Loss:

\text{Log Loss} = -\frac{1}{N} \sum_{i=1}^N y_i \log(\hat{y}_i) + (1 - y_i) \log(1 - \hat{y}_i)

Cross-Validation:
- KFold: Splits data into k folds.
- StratifiedKFold: Preserves class distribution in folds.
Learning Curves:
- Plot training/error vs. iterations/epochs to diagnose overfitting/underfitting.

C. Unsupervised Learning

1. Clustering (K-Means)

Parameters:
- n_clusters: Number of clusters (k).
- init: Initialization method ('k-means++' or 'random').
- max_iter: Maximum iterations.
Evaluation:
- Elbow Method: Plot inertia (within-cluster sum of squares) vs. k.
- Silhouette Score: Measures cluster cohesion/separation.

\text{Silhouette Score} = \frac{b - a}{\max(a, b)}

a: Mean distance to points in the same cluster.
b: Mean distance to points in the nearest other cluster.

2. Dimensionality Reduction

PCA (Principal Component Analysis):
- Transforms features into orthogonal components.
- n_components: Number of principal components to keep.

D. Hyperparameter Tuning

1. Grid Search (`GridSearchCV`)

Exhaustively searches over a parameter grid.
cv: Number of cross-validation folds.
scoring: Metric to optimize (e.g., 'accuracy', 'neg_mean_squared_error').

2. Randomized Search (`RandomizedSearchCV`)

Samples a fixed number of parameter combinations (n_iter).
More efficient for large parameter spaces.

E. Incremental Learning

Models Supporting partial_fit:
- SGDClassifier, SGDRegressor
- GaussianNB
- PassiveAggressiveClassifier

Use Case: Large datasets that don’t fit in memory.

Example:

for chunk in pd.read_csv("large_file.csv", chunksize=1000):
    model.partial_fit(chunk[X_cols], chunk[y_col])

F. Feature Selection

VarianceThreshold:
- Removes features with variance < threshold.
- Example: VarianceThreshold(threshold=0.1).

2. Key Equations

Concept	Equation
Linear Regression	$y = \beta_0 + \beta_1 x_1 + \dots + \beta_n x_n + \epsilon$
Logistic Regression	$\sigma(z) = \frac{1}{1 + e^{-z}}$
MSE	$\text{MSE} = \frac{1}{n} \sum (y_i - \hat{y}_i)^2$
R² Score	$R^2 = 1 - \frac{\text{SS}_{\text{res}}}{\text{SS}_{\text{tot}}}$
Standard Scaling	$z = \frac{x - \mu}{\sigma}$
MinMax Scaling	$x_{\text{scaled}} = \frac{x - x_{\text{min}}}{x_{\text{max}} - x_{\text{min}}}$
Log Loss	$-\frac{1}{N} \sum y_i \log(\hat{y}_i) + (1 - y_i) \log(1 - \hat{y}_i)$
Silhouette Score	$\frac{b - a}{\max(a, b)}$
Gini Impurity	$G = 1 - \sum p_i^2$
Entropy	$H = -\sum p_i \log(p_i)$

Graph 2: Supervised Learning Models

Graph 3: Model Evaluation

Graph 4: Unsupervised Learning

Graph 5: Hyperparameter Tuning

Graph 6: Incremental Learning

4. Exam Tips

Pandas/NumPy:
- Know how to filter (df[df['Salary'] > 50000]), sort (sort_values), and aggregate (groupby, mean).
- Understand dropna and fillna.
Scikit-Learn:
- Memorize key methods: fit, predict, transform, fit_transform, score.
- Know which models support partial_fit (e.g., SGDClassifier, GaussianNB).
Decision Trees:
- Understand how max_depth, min_samples_split, and criterion affect the tree.
- Calculate max leaf nodes: $2^{\text{max\_depth}}$ .
Evaluation Metrics:
- For regression: MSE, R².
- For classification: accuracy, precision, recall, F1, ROC-AUC, log loss.
Clustering:
- Use the elbow method or silhouette score to choose k.
- Know that K-Means requires predefined n_clusters.
Hyperparameter Tuning:
- GridSearchCV vs. RandomizedSearchCV: trade-off between exhaustiveness and efficiency.
Neural Networks:
- hidden_layer_sizes=(5, 3) means 2 hidden layers with 5 and 3 neurons.
- alpha controls regularization (higher = less overfitting).

IIT M IMPROVEMENT FN EXAM QDD1 22 Dec - 22 Dec 2024

Answer key

Concepts this paper tests

1. Core Topics & Concepts

A. Data Preprocessing

Key Concepts:

Key Pandas/NumPy Commands:

B. Supervised Learning

1. Linear Models

2. Decision Trees

3. Ensemble Methods

4. Neural Networks (MLPClassifier/MLPRegressor)

5. Model Evaluation

C. Unsupervised Learning

1. Clustering (K-Means)

2. Dimensionality Reduction

D. Hyperparameter Tuning

1. Grid Search (GridSearchCV)

2. Randomized Search (RandomizedSearchCV)

E. Incremental Learning

F. Feature Selection

2. Key Equations

Graph 2: Supervised Learning Models

Graph 3: Model Evaluation

Graph 4: Unsupervised Learning

Graph 5: Hyperparameter Tuning

Graph 6: Incremental Learning

4. Exam Tips

4. Neural Networks (`MLPClassifier`/`MLPRegressor`)

1. Grid Search (`GridSearchCV`)

2. Randomized Search (`RandomizedSearchCV`)