Download Certified Machine Learning (Python) Exam and more Exams Technology in PDF only on Docsity!
Certified Machine Learning (Python) Practice Exam
Question 1: What is the primary goal of machine learning? Options: A. To design explicit algorithms B. To learn patterns from data C. To compute statistics manually D. To implement database queries Answer: B Explanation: Machine learning focuses on automatically learning patterns from data to make predictions or decisions without explicit programming. Question 2: Which of the following best describes supervised learning? Options: A. Learning from unlabeled data B. Learning from labeled data C. Learning without any feedback D. Learning by trial and error Answer: B Explanation: Supervised learning uses labeled data to train models so that they can predict outcomes for new, unseen data. Question 3: In unsupervised learning, what is the main goal of clustering? Options: A. To predict target values B. To reduce data dimensionality C. To group similar data points D. To enhance image resolution Answer: C Explanation: Clustering aims to group similar data points together based on features and similarities without prior labeling. Question 4: Which Python library is most commonly used for numerical computations in machine learning? Options: A. pandas B. NumPy C. matplotlib D. TensorFlow Answer: B Explanation: NumPy provides support for large, multi-dimensional arrays and matrices, making it essential for numerical computations. Question 5: What does the term “feature scaling” refer to? Options: A. Increasing the number of features
B. Reducing the number of observations C. Normalizing data values to a common scale D. Encoding categorical variables Answer: C Explanation: Feature scaling normalizes data values so that features contribute equally to the model’s performance. Question 6: Which activation function is most commonly used in deep learning hidden layers? Options: A. Softmax B. Sigmoid C. ReLU D. Linear Answer: C Explanation: ReLU (Rectified Linear Unit) is popular because it helps mitigate the vanishing gradient problem while being computationally efficient. Question 7: What is overfitting in machine learning models? Options: A. Underestimating the model’s complexity B. When a model learns noise in the training data C. Having too few training samples D. When a model performs equally on training and test data Answer: B Explanation: Overfitting occurs when a model learns the training data—including its noise—instead of the underlying pattern, resulting in poor generalization. Question 8: Which technique is used for reducing overfitting in neural networks? Options: A. Increasing learning rate B. Dropout C. Using more layers D. Removing bias Answer: B Explanation: Dropout randomly disables neurons during training, which helps prevent the network from overfitting. Question 9: In a confusion matrix, what does the term “True Positive” (TP) represent? Options: A. Incorrectly predicted positive cases B. Correctly predicted negative cases C. Correctly predicted positive cases D. Incorrectly predicted negative cases Answer: C Explanation: True Positives are cases where the model correctly predicts the positive class.
D. Matplotlib Answer: A Explanation: pandas is a powerful library used for data manipulation and analysis, offering data structures like DataFrames. Question 15: In the context of decision trees, what is “pruning”? Options: A. Adding more branches to the tree B. Reducing the depth of the tree to prevent overfitting C. Increasing the number of leaves D. Scaling features Answer: B Explanation: Pruning is the process of reducing the size of a decision tree to improve its generalization by removing branches that have little importance. Question 16: What is the purpose of one-hot encoding in data preprocessing? Options: A. To scale numeric features B. To convert categorical variables into binary vectors C. To impute missing values D. To reduce dimensionality Answer: B Explanation: One-hot encoding transforms categorical variables into a binary matrix representation, which is more suitable for ML algorithms. Question 17: Which metric is most appropriate for evaluating a regression model? Options: A. Accuracy B. Precision C. Mean Absolute Error (MAE) D. F1-Score Answer: C Explanation: Mean Absolute Error (MAE) is commonly used to evaluate regression models by measuring the average absolute differences between predicted and actual values. Question 18: Which of the following is an ensemble learning method? Options: A. Logistic Regression B. k-Nearest Neighbors C. Random Forest D. Support Vector Machine Answer: C Explanation: Random Forest is an ensemble learning method that combines multiple decision trees to improve model accuracy and reduce overfitting. Question 19: In support vector machines, what does the “kernel trick” enable? Options:
A. Faster computation of decision trees B. Transformation of data into higher dimensions C. Simplification of linear models D. Removal of outliers Answer: B Explanation: The kernel trick allows SVMs to operate in a high-dimensional space without explicitly computing the coordinates in that space. Question 20: What is the key concept behind gradient descent? Options: A. Finding the maximum value of a function B. Iteratively updating parameters to minimize the cost function C. Randomly selecting model parameters D. Performing matrix inversion Answer: B Explanation: Gradient descent is an iterative optimization algorithm used to minimize a cost function by updating parameters in the opposite direction of the gradient. Question 21: Which Python library is primarily used for creating visualizations in machine learning? Options: A. pandas B. NumPy C. Matplotlib D. scikit-learn Answer: C Explanation: Matplotlib is widely used in Python for creating a variety of visualizations, including plots and graphs. Question 22: What is the main purpose of data normalization? Options: A. To reduce the number of features B. To standardize the range of independent variables C. To encode categorical features D. To split data into training and testing sets Answer: B Explanation: Normalization scales the data to a specific range, often between 0 and 1, ensuring that each feature contributes equally. Question 23: In Python, which data structure is immutable? Options: A. List B. Tuple C. Dictionary D. Set Answer: B Explanation: Tuples are immutable, meaning their elements cannot be changed after creation.
Answer: A Explanation: A confusion matrix provides a summary of prediction results by comparing predicted and actual classifications. Question 29: What is the main advantage of using ensemble methods like boosting? Options: A. They always reduce the training time B. They combine multiple models to improve accuracy C. They eliminate the need for hyperparameter tuning D. They automatically remove irrelevant features Answer: B Explanation: Ensemble methods improve performance by combining multiple models to reduce bias and variance. Question 30: Which Python library is used for data manipulation with DataFrames? Options: A. NumPy B. pandas C. scikit-learn D. TensorFlow Answer: B Explanation: pandas provides powerful data structures like DataFrames for handling and analyzing structured data. Question 31: What is the primary purpose of exploratory data analysis (EDA)? Options: A. To build the final model B. To understand the data’s characteristics C. To deploy machine learning models D. To perform hyperparameter tuning Answer: B Explanation: EDA is used to analyze data sets to summarize their main characteristics, often visualizing data to uncover patterns and anomalies. Question 32: Which technique is used to reduce the dimensionality of a dataset while preserving variance? Options: A. k-NN B. PCA C. SVM D. Random Forest Answer: B Explanation: Principal Component Analysis (PCA) reduces dimensionality by projecting data onto principal components that capture the maximum variance. Question 33: In Python, which structure allows you to store key-value pairs? Options:
A. List B. Tuple C. Dictionary D. Set Answer: C Explanation: Dictionaries in Python store data in key-value pairs and provide fast lookup capabilities. Question 34: What does the term “epoch” refer to in neural network training? Options: A. A single pass through the entire training dataset B. A single training sample C. The initial weight configuration D. The final evaluation phase Answer: A Explanation: An epoch is one complete pass through the entire training dataset during the training process. Question 35: Which metric is used to evaluate classification models by balancing precision and recall? Options: A. Mean Squared Error B. R-squared C. F1-Score D. Adjusted R-squared Answer: C Explanation: The F1-Score is the harmonic mean of precision and recall, providing a balance between the two metrics for classification performance. Question 36: Which Python library provides support for machine learning algorithms and tools? Options: A. scikit-learn B. pandas C. NumPy D. Matplotlib Answer: A Explanation: scikit-learn is a widely used library offering simple and efficient tools for data mining and data analysis. Question 37: What is the main concept behind the bagging technique in ensemble methods? Options: A. Combining predictions by averaging multiple models trained on random subsets B. Sequentially adding models to correct errors C. Using a single model with multiple outputs D. Dividing data into clusters Answer: A Explanation: Bagging (Bootstrap Aggregating) trains multiple models on random subsets of the data and averages their predictions to reduce variance.
D. Hyperparameter tuning Answer: C Explanation: Data preprocessing involves cleaning data by handling missing values, outliers, and inconsistencies before model training. Question 43: What is the purpose of the “learning rate” in gradient descent? Options: A. To determine the number of epochs B. To control the step size in updating parameters C. To decide the number of features D. To measure the model’s accuracy Answer: B Explanation: The learning rate determines how much to adjust model parameters during each iteration of gradient descent, impacting convergence speed and stability. Question 44: Which algorithm is best suited for predicting continuous outcomes? Options: A. Logistic Regression B. Linear Regression C. Decision Trees D. Naive Bayes Answer: B Explanation: Linear regression is used to predict continuous outcomes by modeling the relationship between variables with a straight line. Question 45: In k-nearest neighbors (k-NN), what does the parameter “k” represent? Options: A. The number of features B. The number of nearest neighbors used for prediction C. The depth of the decision tree D. The number of clusters Answer: B Explanation: In k-NN, “k” is the number of nearest neighbors that are considered when making a prediction about a data point. Question 46: Which evaluation metric is particularly useful when dealing with imbalanced classification datasets? Options: A. Accuracy B. F1-Score C. Mean Squared Error D. R-squared Answer: B Explanation: The F1-Score balances precision and recall, making it more informative than accuracy when classes are imbalanced.
Question 47: What does the term “cross-validation” help to achieve? Options: A. Faster model training B. Reliable estimation of model performance C. Data normalization D. Feature extraction Answer: B Explanation: Cross-validation helps to reliably estimate model performance by repeatedly splitting data into training and testing sets. Question 48: Which machine learning algorithm is based on Bayes’ theorem and assumes independence among predictors? Options: A. Logistic Regression B. Decision Trees C. Naive Bayes D. Support Vector Machine Answer: C Explanation: Naive Bayes classifiers use Bayes’ theorem and assume that predictors are independent, simplifying computations. Question 49: What is the primary benefit of using GPUs in machine learning? Options: A. They reduce data preprocessing time B. They accelerate the training of complex models C. They simplify model interpretation D. They replace the need for CPUs Answer: B Explanation: GPUs can handle parallel computations efficiently, significantly speeding up the training of complex models, especially deep neural networks. Question 50: What is the significance of the “activation function” in a neural network? Options: A. It determines the number of layers B. It introduces non-linearity into the model C. It splits the dataset D. It scales input features Answer: B Explanation: Activation functions introduce non-linearity into neural networks, enabling them to learn complex patterns. Question 51: Which Python data structure would be most suitable for ordered collections that can be modified? Options: A. Tuple B. Set
Question 56: What does “recursive feature elimination” (RFE) accomplish? Options: A. It removes redundant observations B. It selects important features by recursively removing less significant ones C. It scales features to a similar range D. It encodes categorical data Answer: B Explanation: RFE is a feature selection technique that recursively eliminates less important features to improve model performance. Question 57: Which method is commonly used to assess the clustering quality of unsupervised algorithms? Options: A. Silhouette Score B. Accuracy C. Confusion Matrix D. F1-Score Answer: A Explanation: The Silhouette Score measures how similar an object is to its own cluster compared to other clusters, evaluating clustering quality. Question 58: In decision trees, what is “information gain”? Options: A. A method for pruning trees B. The decrease in entropy after a dataset is split C. A technique for scaling data D. A measure of data correlation Answer: B Explanation: Information gain quantifies the reduction in entropy after splitting a dataset on a particular feature. Question 59: Which library would you use for building a convolutional neural network (CNN) in Python? Options: A. scikit-learn B. TensorFlow/Keras C. pandas D. NumPy Answer: B Explanation: TensorFlow (often with the Keras API) is widely used for constructing deep learning models such as convolutional neural networks. Question 60: Which concept refers to the trade-off between a model’s complexity and its performance on new data? Options: A. Bias-Variance Trade-off
B. Data augmentation C. Feature scaling D. Model stacking Answer: A Explanation: The bias-variance trade-off describes the balance between the error introduced by overly simplistic models (bias) and the error from overly complex models (variance). Question 61: What does “object-oriented programming” (OOP) in Python primarily involve? Options: A. Writing functions only B. Organizing code into objects and classes C. Using only built-in data types D. Scripting without any functions Answer: B Explanation: OOP involves structuring code into classes and objects, enabling code reuse, encapsulation, and inheritance. Question 62: What is the purpose of a “heatmap” in exploratory data analysis? Options: A. To show the distribution of a single variable B. To display correlations between variables C. To split the dataset into training and testing sets D. To encode categorical data Answer: B Explanation: Heatmaps visually represent correlation matrices, making it easy to identify strong relationships between variables. Question 63: Which algorithm is best known for its simplicity and instance-based learning? Options: A. Naive Bayes B. k-Nearest Neighbors (k-NN) C. Decision Trees D. Support Vector Machines Answer: B Explanation: k-NN is an instance-based algorithm that makes predictions based on the similarity between new data points and existing examples. Question 64: What is a common use case for autoencoders in machine learning? Options: A. Data compression and anomaly detection B. Data splitting C. Feature scaling D. Hyperparameter tuning Answer: A Explanation: Autoencoders compress data into lower-dimensional representations and are often used for anomaly detection.
C. do-while loop D. recursive loop Answer: B Explanation: A for loop in Python is typically used when iterating over a sequence with a known number of iterations. Question 70: What does “SMOTE” stand for in the context of handling imbalanced data? Options: A. Synthetic Minority Over-sampling Technique B. Simple Majority Over-sampling Technique C. Standardized Minority Oversampling Engine D. Scalable Model Over-sampling Tool Answer: A Explanation: SMOTE stands for Synthetic Minority Over-sampling Technique, a method to generate synthetic samples for the minority class. Question 71: Which type of neural network is best suited for processing sequential data? Options: A. Convolutional Neural Network (CNN) B. Recurrent Neural Network (RNN) C. Decision Tree D. Random Forest Answer: B Explanation: RNNs are designed to handle sequential data by maintaining internal memory to capture context over time. Question 72: Which evaluation metric is best suited for multi-class classification problems? Options: A. Binary cross-entropy B. Categorical cross-entropy C. Mean Squared Error D. Adjusted R-squared Answer: B Explanation: Categorical cross-entropy is used when evaluating multi-class classification models as it measures the performance of a classification model whose output is a probability value between 0 and
Question 73: In Python’s pandas library, which function is used to read CSV files? Options: A. read_excel() B. read_json() C. read_csv() D. load_csv() Answer: C Explanation: The read_csv() function in pandas is used to load CSV files into a DataFrame for further analysis.
Question 74: What is the primary difference between artificial intelligence (AI) and machine learning (ML)? Options: A. AI is a subset of ML B. ML is a subset of AI C. They are exactly the same D. AI only deals with robotics Answer: B Explanation: Machine learning is a subset of artificial intelligence that focuses specifically on learning patterns from data. Question 75: What does the “test set” in machine learning represent? Options: A. The data used to train the model B. The data used to validate the model during training C. The data used to evaluate model performance after training D. The data used for feature engineering Answer: C Explanation: The test set is used only after model training to evaluate how well the model generalizes to unseen data. Question 76: Which algorithm is typically used for text classification tasks? Options: A. Linear Regression B. Naive Bayes C. k-Means Clustering D. PCA Answer: B Explanation: Naive Bayes classifiers are popular for text classification due to their simplicity and effectiveness, especially when dealing with high-dimensional data. Question 77: In Python, what is the primary purpose of the “pandas DataFrame”? Options: A. To perform complex mathematical operations B. To store and manipulate tabular data C. To create visualizations D. To handle deep learning models Answer: B Explanation: A DataFrame is a two-dimensional data structure used in pandas to store and manipulate tabular data efficiently. Question 78: Which loss function is typically used for multi-class classification in neural networks? Options: A. Mean Squared Error B. Categorical Cross-Entropy C. Hinge Loss
Question 83: Which machine learning algorithm is inherently based on distance metrics? Options: A. Logistic Regression B. k-Nearest Neighbors C. Decision Trees D. Naive Bayes Answer: B Explanation: k-Nearest Neighbors (k-NN) relies on distance metrics (such as Euclidean distance) to determine the similarity between data points. Question 84: What does “hyperparameter tuning” refer to? Options: A. Training the model with the default parameters B. Adjusting the parameters that govern the learning process C. Scaling the input features D. Evaluating model performance Answer: B Explanation: Hyperparameter tuning involves adjusting the parameters that control the learning process (such as learning rate or tree depth) to optimize model performance. Question 85: Which technique is commonly used to visualize the distribution of a single variable? Options: A. Box plot B. Scatter plot C. Heatmap D. Bar chart Answer: A Explanation: A box plot is used to visualize the distribution, central value, and variability of a single variable. Question 86: What is the role of the “validation set” in machine learning? Options: A. It is used to train the model B. It is used to fine-tune model hyperparameters C. It is used to test the final model D. It is used for feature scaling Answer: B Explanation: The validation set is used during training to fine-tune hyperparameters and make decisions about the model architecture. Question 87: Which algorithm is most appropriate for anomaly detection in high-dimensional data? Options: A. k-Means Clustering B. Autoencoders C. Linear Regression D. Decision Trees
Answer: B Explanation: Autoencoders can learn efficient data representations and are effective for detecting anomalies by reconstructing inputs. Question 88: What is the purpose of dimensionality reduction in machine learning? Options: A. To increase model complexity B. To reduce the number of random features C. To remove redundant features and simplify data D. To encode categorical data Answer: C Explanation: Dimensionality reduction techniques like PCA reduce the number of features by removing redundancy, leading to simpler and faster models. Question 89: Which of the following is NOT a type of machine learning? Options: A. Supervised Learning B. Unsupervised Learning C. Reinforcement Learning D. Reactive Learning Answer: D Explanation: Reactive Learning is not a recognized type of machine learning; the primary types include supervised, unsupervised, and reinforcement learning. Question 90: In deep learning, what is “backpropagation” used for? Options: A. Propagating inputs forward through the network B. Calculating gradients to update weights C. Normalizing data inputs D. Segmenting images Answer: B Explanation: Backpropagation calculates the gradient of the loss function with respect to each weight, allowing the network to update weights appropriately during training. Question 91: Which machine learning model is best suited for handling linear relationships between variables? Options: A. Linear Regression B. Decision Trees C. k-NN D. SVM with RBF kernel Answer: A Explanation: Linear regression is specifically designed to model linear relationships between independent and dependent variables. Question 92: Which of the following best describes “data cleaning”? Options:
