Analyze and interpret output from models

Reference no: EM133096400

Objective

The objective of this exercise is to generate association rules (or affinity) for the survivability of the passengers on RMS Titanic.

Activities

• Import and prepare data
• Apply data mining algorithms
• Configure predictive models
• Create data visualizations
• Analyze and interpret output from models

SCENARIO

Using an Excel data file containing information about the passengers of the Titanic, you will use association analysis to generate rules for their survivability.

ASSOCIATION ANALYSIS

Several predictive models are available in SAP Predictive Analytics. More can be integrated from the R language. We would like to discover the associations among items. These are presented as rules with values for support, confidence, and lift for each rule

1. We will now do an association analysis (using an Apriori algorithm) for the passenger data in the Titanic disaster 2
a. Launch SAP Predictive Analytics

b. Click on Expert Analytics, then on Expert Analytics.

c. Create a new document. Choose MS Excel as Data Source. Next.

d. Browse for the titanic_E11_1.xlsx file. Create.

2. We will now launch the prediction capabilities of SAP Expert Analytics

a. Click on Predict. You are now in the Designer tab.

b. You will see several Algorithms such as Regression, Outliers, Time Series, Decision Trees, Neural Network, Clustering and Association. See Figure 1.

c. And you see the data source titanic_E11_1.xlsx.

d. Double-click the R-Apriori algorithm. The algorithm is automatically connected to the data source

e. Roll your mouse over the algorithm and click on Configure Settings

f. Item Column(s) - Select Class, Sex, Age and Survived

g. Support: 0.01, (leave confidence at .8)

h. Done

i. Click Run

3. The algorithm is now generating the association rules. After the execution is complete, click OK to review the results.
a. You see a table of rules that were generated.

b. Now let's rerun the algorithm so that only Survived is on the right hand side of the
rules.

4. Go to Designer tab (on the right)

a. Edit the R-Apriori properties by selecting configure settings

b. Click on Advanced tab.

c. In Rhs Item(s) type: Survived=No,Survived=Yes (type without spaces in between)

d. Choose Default Appearance: Lhs Items

e. In the Performance tab, select Sort Type: Descending

f. Done. Run the analysis again.

g. View the results

h. You now see the results for all the Rhs (right-hand side or consequent) for Survived (No, Yes). See Figure 2.

i. Click on Association Chart. Here you can see the results in a tag cloud format.

5. Click on Visualize

a. Select component R-Apriori

b. Convert the attributes Confidence , Support, and Lift to Measures (by right clicking on them and selecting ‘create a measure')
c. Change each measure's aggregation to None (from the default Sum). You may also wish to rename each measure.
d. Create a bubble chart (available under scatter plots). X-Axis - Support, Y Axis -
Confidence, Bubble width - Lift

e. Add the Rules from Attributes to Dimensions: Legend Color

f. You can now see the large bubbles indicating the lift for that rule. Lift indicates the

strength of a rule over the random co-occurrence of the independent and the dependent variables, given their individual support.
6. We can now export the results of our association analysis

a. Go to the predict tab

b. Click on Designer tab

c. Click on Data Writers drop down list

d. Add a CSV writer to our analysis by double clicking on the CSV Writer menu.

e. Edit its properties by selecting configure settings ? properties

f. Choose a File name and type by clicking on Browse. .csv is the default file type.

g. Save and Close

h. Run the CSV writer

i. You can open the CSV file that was generated to review the results

Question 1: What is meant by support, confidence and lift?

Question 2: Which rule is most dependable within the rules you have found? Why? Question 3: Why did you set a filter (see 4.c) on the consequent in the rules?

Attachment:- Configure predictive models.rar

Reference no: EM133096400

Questions Cloud

Understanding of ethics : You are the payroll manager at Widgets R Us. You need to calculate an employee's gross and net pay based on the table below. The work week for this company is 4

How much will the williams still owe on the house : Their monthly payments are $1346.31. After making their first payment, how much will the Williams still owe on the house

Areas of organization in operational and strategic : How HR link with areas of organization in operational and strategic way

Failure of the performance appraisal : Alam Bina Sdn. Bhd.'s employees are unhappy with the performance appraisal conducted year after year. As a new Human Resource Senior Manager of the company

Analyze and interpret output from models : The objective of this exercise is to generate association rules (or affinity) for the survivability of the passengers on RMS Titanic

Client expectation of the counselling process : Identify at least four (4) skills you would use to clarify, confirm or modify a client's expectation of the counselling process.

Identify at least six client issues : a. Identify at least six (6) client issues that may be beyond your role as a counsellor.

What was the conversion cost for Valencia Orange : Question - Valencia Orange manufactures orange juice. What was the conversion cost for Valencia Orange's Jacksonville operation last month

Create a counselling plan : a. Briefly explain what you consider regarding Kirstie's limited English when providing her with information about the counselling service.

User Account

All Pages