Assignment title: Information
MIS 4380/MBA 5380 Business Intelligence
Titanic Assignment Feedback
You are expected to use the CRISP-DM process model to organize your report. CRISP-DM has 6 steps and several of them are regularly being skipped. You should review the coverage of CRISP-DM in the Sharda text for guidance. For the Titanic assignment, I would expect the steps to address the following (although in much greater detail and certainly not exclusively):
1. Business/Organizational Understanding:
The business involved is a cruise line. Why would that business want to conduct this analysis? What do they get from the results? (I know the problem wasn't framed this way, but that is what the Business Understanding is about.)
2. Data Understanding:
Where does the data come from? What attributes does it consist of? Are there any apparent problems in the raw data? Can you do some simple graphs or statistical calculations to learn more about the data?
3. Data Preparation:
Is there missing data? If so, how was it handled.
Does any data need to be converted into another form? If so, what will be done with it. Do you have to eliminate any attributes? Will you exclude some attributes in the modeling process? Which ones and why?
What other data cleaning actions were necessary?
4. Model Building:
What modeling technique is being used and why? ("decision tree, because that is what was assigned" is not good enough.) Describe the steps you took to build your final model and include supporting screen shots.
5. Evaluation:
Does the final model allow you to meet the objectives set forth in step 1?
How accurate does the model appear to be?
In the case of the Titanic data, we test the model using family data. What are the results? What do they mean?
6. Deployment:
How will the model be used?
You will probably also want a final section to address your conclusions and include any suggestion(s) of what should be done next to address the problem/issue.
You should apply these same concepts to the comprehensive problem scenario.