Supervised Learning Quiz

 Supervised Learning Quiz

1. Regression Modeling Quiz

Ques 1: Logistic Regression requires that data meet certain Assumptions

Answer: FALSE

Feedback

Linear regression, not logistic regression requires that the data it's used on meet certain assumptions. It assumes a linear relationship between continuous variables.

Ques 2: What is NOT a typical task in Data Cleaning?

Answer: predicting outcomes

Ques 3: What sample of data should you use to assess if your Machine Learning Model has been trained properly?

Answer: testing set

Feedback

Spot on! The testing set is used to first measure how well a model learns from the training set and the validation set is used to further tune model hyperparameters.

Ques 4: Logistic Regression models can predict numerical and categorical outcomes.

Answer: TRUE

You got it! While linear regression can only be used to predict continuous, numerical outcomes, logistic regression can predict both categorical and numerical responses.

Ques 5: What does a model's R squared represent?

Answer: the percentage variation in the dependent variable that is explained by the independent variable,

• 2. Decision Tree Quiz

• Ques 1: What is NOT a metric for evaluating classification problems?

• (a) precision

• (b)  F1 Score

• (c) Root Mean Squared Error

• (d) accuracy

• Answer: Option C: Root Mean Squared Error

• Feedback

•  Root Mean Square Error is used to evaluate regression problems!

• 
  

• Ques 2: Which of these represents the bagging method for random forests?

• (a) It estimates statistics on a population by sampling a dataset with replacement.

• (b)  A dataset is partitioned into k groups, where each group is given the opportunity of being used.

• (c)  It creates a collection of predictors to reduce variance.

• (d) Each observation has the same chance of appearing in the subset.

• Answer: Option D: Each observation has the same chance of appearing in the subset.

• 
  

• Ques 3: Which of the following is a hyperparameter you can tune for Decision trees?

(a)  Batch Size

(b)  Learning Rate

(c) Root Node

(d)  Split Criteria

Answer: Option(D): Split

Ques 4: Decision Trees are prone to _____.

(a) little interpretability

(b)  ethical problems

(c)   not converging

(d)   overfitting

      Answer: Option(D): overfitting

    Ques 5: Which of the following is an algorithm used to split decision trees?

·    (a) Variance

·    (b) Random Forests

·    (c) Support Vector Machines

·    (d) Gini index

      Answer: Option(D): Gini index

0 Ques 6: Which of these is NOT a decision tree pruning strategy?

·   (a) minimum error

·   (b) information entropy

·   (c) smallest tree method

·    (d) early stopping

Answer: Option(A): minimum error

3. K-Nearest Neighbors Quiz

Ques 1: Which evaluation metric takes the mean of precision and recall to provide a better overall measure of model's performance?

(a)  ROC-Curve

·    (b) sensitivity

     (c)  F1 Score

·    (d)  accuracy

    

    Answer: Option(C): F1 Score

    Ques 2: Which of these is a method of finding the best value for K?

     (a) Depth-First Search

·    (b) Uniform Cost Search

·    (c) A * search

·    (d) Grid Search

    Answer: Option(D): Grid Search

    Ques 3: Which of these is the first step in building a KNN model?

·   (a)  Measure distance between data points.

·   (b)  Predict where data points should be.

·   (c)  Determine a Distance Metric.

·   (d)  Plot all data points.

        

       Answer: Option(A): Measure distance between data points.

    Ques 4: If you have 2 million data points, KNNs be a computationally cheap model.

·      (a)   TRUE

·      (b)   FALSE

    

    Answer: Option(B): FALSE  

    Ques 5: What is NOT a common distance measure in KNN algorithms?

·   (a) Hamming

·   (b)  Franciscan

·   (c)  Manhattan

·   (d)  Euclidean

Answer: Option(B): Franciscan

Feedback

This isn't a type of distance measure.

Ques 6: You must tune hyperparameters when building KNN Models.

·   (a)  TRUE

·   (b)  FALSE

    

    Answer: Option(A): TRUE

4. 4. Neural Network Quiz

    Ques 1: Which of these is a type of Artificial Neural Network?

·   (a) Support Vector Machines

·   (b) Recurrent Neural Networks

·   (c) Ensemble Methods

·   (d) Biological neural networks

    

    Answer: Option(B): Recurrent Neural Networks

    Ques 2: Which of these is NOT a kind of activation function.

·   (a) ReLU

·   (b) Specificity

·   (c) TanH

·   (d) Sigmoid

 Answer:Option(B): Specificity

Ques 3: Which of these is NOT a layer in the artificial neural network structure?

·  (a)  Hidden Layer

·  (b)  Middle Layer

·  (c)  Output Layer

·   (d)  Input Layer

    Answer: Option(B): Middle Layer

    Ques 4: Why is scaling numerical features important for neural networks?

·  (a)  In order to avoid overfitting

·  (b) To make models easier to read

·  (c)  To reduce the training time

·  (d)  To allow the network to make accurate predictions

    

    Answer: Option(D): To allow the network to make accurate predictions

    Ques 5: AUC a metric used to evaluate neural network models.

·  (a) TRUE

·   (b) FALSE

    

    Answer: Option(A): TRUE

    

    Ques 6: Which step is most crucial to model reproducibility?

·  (a) Documentation

·  (b) Stratified Sampling

·  (c) Model Training

·  (d) Specificity

Answer: Option(A) Documentation