ANL305 TMA (Tutor-Marked Assignment) SUSS : July 2024 – Association and Clustering, Singapore

Question 1

Based on the given background and dataset details, identify a potential business problem. State the business objective and data mining objective and appraise the suitability of using association rule mining for solving the proposed business problem.

Question 2

Import the data into the SPSS Modeler and construct Apriori models using the following settings (Data preparation is NOT required for Q2 before performing the following data and model settings):

• Data Node:

o Measurement setting: click “Read Values” under the tab “Types”. Keep the default “Measurement” setting unchanged.

o Role setting: Nobeyesdad = Target, non-Continuous fields = Input, Continuous fields = None

• Apriori Node:

o Parameter setting 1: default setting

o Parameter setting 2: your own customized setting

Answer the following questions:

a) For parameter setting 1, provide a screenshot of the generated rules, including the total number of rules and the first three rules. Discuss the modeling results and findings using parameter setting 1, with reference to the business problem you have proposed in Q1.

b) Describe your own parameter setting 2 in detail and provide a screenshot of the generated rules to show the total number of rules and the first three rules. Compare the generated rules with those from parameter setting 1 and discuss any new insights or findings.

Question 3

Suppose you would like to explore the numeric fields of the dataset and then construct other model(s), answer the following questions to complete the work:

(a) Import the data file into a new data node and apply the following Data Node settings:

• Keep the default measurement setting unchanged.
• Set the Role of Nobeyesdad = Target and all other fields = Input.

Report distribution screenshots and use up to 50 words to describe the distributions of the three numeric fields: Age, Height, Weight.

(b) Transform all the three numeric fields into categorical variables using the “Binning” node in the SPSS Modeler. You can select the “Equal-width” method or the “Tiles (Equalcount)” method to complete the binning work. Report the distribution of the transformed fields.

Note: Each of the newly generated fields should have no more than four categories (aka bins). You may use the same binning method (or a mixture of different binning methods) for the fields. It is also acceptable to propose an alternative binning method if you think the binning methods provided by the IBM SPSS Modeler are not suitable for some fields. Justify your choices briefly

(c) Link a “Type” Node to the “Binning” node and report the screenshot of role setting. Construct new Apriori model(s) using the transformed dataset. Analyze and discuss your modeling results, including:

• Apriori node setting
• The number of generated rules
• Important rules
• A summary of the findings
• A discussion on the impacts of the three new fields, comparing these to the findings in Q2

(d) Propose two (2) deployment suggestions to answer the business problem proposed in Q1. Discuss the potential limitations of the study and propose one improvement.