New Semester
Started
Get
50% OFF
Study Help!
--h --m --s
Claim Now
Question Answers
Textbooks
Find textbooks, questions and answers
Oops, something went wrong!
Change your search query and then try again
S
Books
FREE
Study Help
Expert Questions
Accounting
General Management
Mathematics
Finance
Organizational Behaviour
Law
Physics
Operating System
Management Leadership
Sociology
Programming
Marketing
Database
Computer Network
Economics
Textbooks Solutions
Accounting
Managerial Accounting
Management Leadership
Cost Accounting
Statistics
Business Law
Corporate Finance
Finance
Economics
Auditing
Tutors
Online Tutors
Find a Tutor
Hire a Tutor
Become a Tutor
AI Tutor
AI Study Planner
NEW
Sell Books
Search
Search
Sign In
Register
study help
business
business analytics data
Data Mining And Predictive Analytics 2nd Edition Daniel T Larose, Chantal D Larose - Solutions
Use the training set to train a CART model, a logistic regression model, and a neural network model to be your set of base classifiers for predicting Churn.
Partition the data set into a training data set and a test data set.
Apply a misclassification cost of 2 (rather than the default of 1) for a false negative. Redo Exercises 23–29 using the new misclassification cost. Make sure to evaluate the models using the new misclassification cost rather than the measures mentioned in Exercise 28.Use the Churn data set to
Evaluate all base classifier models and all voting ensemble models with respect to overall error rate, sensitivity, specificity, proportion of false positives, and proportion of false negatives. Which model performed the best?
Combine the classification results into voting ensemble models, using the following methods:a. Majority classificationb. Single sufficient classificationc. Twofold sufficient classificationd. Positive unanimity classification.
Apply the base classifier models to the test data set.
Use the training set to train a CART model, a logistic regression model, and a neural network model to be your set of base classifiers for predicting Income.
True or false: Ensemble models using voting or propensity averaging do not perform well with misclassification costs.
Describe how propensity averaging ensemble models would behave, for the following:a. Lower threshold valuesb. Higher threshold values.
How does a threshold value of t define positive and negative responses of the target variable?
When scanning the normalized histogram of mean propensity values, what should we look for in a candidate threshold value?
True or false: Propensity is a characteristic of a data set rather than a single record.
For an ensemble of m base classifiers, state in words the formula for mean propensity.
For a binary target, how is the propensity for a positive response calculated?
What is the rationale for using propensity averaging rather than a voting ensemble?
True or false: Voting ensemble models always perform better than any of their constituent classifiers.
Is a voting ensemble model constructed from the classification results of the training set or the test set?
What is a detriment of using voting ensemble models?
Describe the characteristics of the models associated with the following voting methods:a. Single sufficient classificationb. Positive unanimity classificationc. Majority classification.
Describe what negative unanimity would be.
Explain what single sufficient and twofold sufficient classification represent.
What is the difference between majority classification and plurality classification?
What is another term for simple model voting?
Calculate the pseudo-F statistic and p-value for each candidate, and select the candidate with the smallest p-value as the best clustering solution
Use a clustering algorithm to develop a clustering solution for a variety of values of k.
With the test data set, apply k-means with the value of k from the preferred model above. Perform validation of the clusters you uncovered with the training and test data sets of the preferred model.
Develop a good classification model for predicting loan approval, based solely on cluster membership. Apply data-driven misclassification costs as shown in Chapter 16. Compare your results for the k = 3 and k = 4 cases using overall model cost. Which model is preferred?
Compare the pseudo-F statistics for the two cluster models. Which model is preferred?
Compare the mean silhouette values for the two cluster models. Which model is preferred?
Repeat Exercises 18–22 using k-means with k = 4.
Using the same variables as the previous exercise, provide a two-dimensional scatter plot, with an overlay of binned silhouette values, as shown in this chapter. Comment on the relationship between your two scatter plots.
Provide a two-dimensional scatter plot, using variables of your choice, with an overlay of cluster membership. Choose variables that result in an interesting plot. Note where the cluster boundaries are close, and where they are not so close.
Calculate the mean silhouette values for each cluster, as well as the overall mean silhouette for the cluster model.
Generate a silhouette plot of your cluster model.
Use k-means with k = 3 to generate a cluster model with the training data set.
What are the criteria for determining whether there is a match between the clusters uncovered in the training and test data sets?
Why might statistical hypothesis tests not be very helpful for big data applications?
What is our cluster validation methodology?
True or false: The best clustering model is the one with the largest value of pseudo-F. Explain.
Explain how we can use the pseudo-F statistic to select the optimal number of clusters.
Why does the pseudo-F statistic have the word pseudo in its name?
Explain how the pseudo-F statistic accounts for both separation and cohesion.
Should the analyst always choose the cluster solution with the better mean silhouette value? Explain.
Describe what a silhouette plot is.
When will a data value have a perfect silhouette value? What is this value?
How is average silhouette interpreted?
Explain how silhouette accounts for both separation and cohesion.
How do we interpret a silhouette value?
What is a silhouette? What is its range? Is it a characteristic of a cluster, a variable, or a data value?
Why is SSE not necessarily a good measure of cluster quality?
What is cluster separation and cluster cohesion?
Why do we need evaluation measures for cluster algorithms?
Briefly profile the clusters for the winning model from the previous exercise.
Calculate model cost for each of the five different sortings. Which model has the highest profitability or the lowest cost?
Run BIRCH on each of the five different sortings. Report the value of k and the MS for each.
Generate four different sortings of the Loans_training data set. Together with the original order from the No Interest model you generated earlier, this makes five different sortings.
Based on your work in the previous exercises, what is the lesson we should learn? For Exercises 13–16, using the Loans data set, demonstrate that different sortings may lead to different numbers of clusters. Make sure you do not include interest as an input to the clustering algorithms.
Evaluate each CART model using the Loans_test data set. Provide contingency tables. Compare the model costs, as in Table 21.4.
Using the Loans_training data set, develop CART models for predicting loan approval, based on cluster membership only, for the two cluster models.
Follow the methodology in Case Study Lesson One to develop cluster models with and without interest.
Is the MS value always indicative of the best cluster solution?
Why is it bad practice to include two highly correlated inputs to a clustering algorithm?
Why is Phase 2 of the BIRCH algorithm efficient?
Describe the parameters of the CF tree.
How are the CFs for two clusters merged?
What is a CF?
Describe the two phases of the BIRCH clustering algorithm.
Why is BIRCH appropriate for streaming data?
Use cluster membership as a further input to a C4.5 decision tree model for classifying income. How important is clustering membership in classifying income? Compare to the CART model.
Use cluster membership as a further input to a CART decision tree model for classifying income. How important is clustering membership in classifying income?
Using the information above and any other information you can bring to bear, construct detailed and informative cluster profiles, complete with titles.
Generate numerical summaries for the clusters. For example, generate a cluster mean summary.
If your software supports this, construct a web graph of income, marital status, and the other categorical variables. Fine-tune the web graph so that it conveys good information.
Construct a bar chart of the cluster membership, with an overlay of marital status. Discuss your findings.
Construct a bar chart of the cluster membership, with an overlay of income. Discuss your findings. Compare to the scatter plot.
Construct a scatter plot (with x/y agitation) of the cluster membership, with an overlay of income. Discuss your findings.
Apply the Kohonen clustering algorithm to the data set, being careful not to include the income field. Use a topology that is not too large, such as 3 × 3.
This chapter shows how cluster membership can be used for downstream modeling. Does this apply to the cluster membership obtained by hierarchical and k-means clustering as well?
Describe what would happen if the learning rate ???? did not decline?
Using weights and distance, explain clearly why a certain output node will win the competition for the input of a certain record.
Describe the three characteristic processes exhibited by SOMs such as Kohonen networks. What differentiates Kohonen networks from other SOM models?
Describe some of the similarities between Kohonen networks and the neural networks of Chapter 7. Describe some of the differences.
Construct and interpret separate ROI charts for the four models. (Extra credit: Find a way to construct a single ROI chart comparing the four models.) Which model is preferred, and why?
Construct and interpret separate profits charts for each of the four models. (Extra credit: Find a way to construct a single profits chart comparing the four models.) Where is the peak profitability for each model? At what percentile does peak profitability occur? Which model is preferred, and why?
Prepare and interpret a response chart comparing four two models.
Construct and interpret a gains chart comparing the four models.
Construct a single lift chart for evaluating the four models: CART, C5.0, neural networks, and logistic regression. Interpret the chart. Which model does better? Is one model uniformly better?
Using the Loans_training data set, construct a neural networks model and a logistic regression model for predicting loan approval, using the rebalanced data.
Neural networks and logistic regression in modeler do not admit explicit misclassification costs. Therefore undertake rebalancing of the data set as a surrogate for the misclassification costs used in this chapter.
Construct and interpret separate ROI charts for the two models. (Extra credit: Find a way to construct a single ROI chart comparing the two models.) Which model is preferred, and why? For Exercises 15–18 we use rebalancing as a surrogate for misclassification costs, in order to add neural
Construct and interpret separate profits charts for the CART model and the C5.0 model. (Extra credit: Find a way to construct a single profits chart comparing the two models.) Where is the peak profitability for each model? At what percentile does peak profitability occur? Which model is preferred,
Prepare and interpret a response chart comparing the two models. Compare the response chart to the lift chart.
Construct and interpret a gains chart comparing the two models.
Construct a single lift chart for evaluating the two models. Interpret the chart. Which model does better? Is one model uniformly better?
Using the Loans_training data set, construct a CART model and a C5.0 model for predicting loan approval.
Should these charts be carried out on the training data set or the test data set? Why?
What is ROI?
Describe what a good profits chart might look like.
Which charts can the analyst use to graphically evaluate the classification models in terms of costs and revenues?
What is a response chart? Which other chart is it similar to?
Showing 2200 - 2300
of 4107
First
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Last
Step by Step Answers