This data set consists of genuine credit records from a South German bank. The aim would generally
Question:
This data set consists of genuine credit records from a South German bank. The aim would generally be to predict which customers will repay the loan in full and which of them will not. There are 1000 records and all amounts are in Deutschmarks. Answer the following using suitable approaches whether descriptive/graphical or inferential and using a suitable package e.g. StatTools. Justify your answers in the main text and include all workings as appendix.
a) Wherever possible and meaningful, provide a brief analysis of each variable, including their distribution, outliers, etc.
b) Does there seem to be differences in age, length of loan, or amount of loan for those who repaid their loans and those who defaulted?
c) Explore and describe the association of each variable with the credit status.
d) Does the Length of the loan vary with the use of the loan?
e) Determine relationships, if any, between Age, Length of loan and Amount of loan.
f) Construct a 3-way contingency table from the factors credit, record and use, and analyse it. You must state your final conclusions in detail.
Problem 2 (30 marks)
John Boris, the founder and major stockholder of DataSoft Inc. has just been told by his lawyer, Michael Grove, that Brexoft Computer Software has informed him through their lawyer that Brexoft plans suing DataSoft for copyright infringements by DataSoft's recent database software Easybase Release 1.1. Brexoft claims damage £250,000 and triple punitive penalties of £750,000. Michael Grove also tells John Boris that Brexoft's lawyer has hinted that would be willing to settle out of court by granting DataSoft retroactively a flat-fee licence for using the software in contention for £1,000,000. This would allow DataSoft to continue selling Release 1.1. John Boris estimates that the future revenue potential for Release 1.1 amounts to roughly £1,400,000. If DataSoft chose to defend the law suit and lost, these sales would also be lost. Michael Grove estimates that the cost of defending the law suit will amount to £250,000. If DataSoft wins, Brexoft will have to refund £250,000. However if Datasoft loses, it will have to refund Brexoft the same amount.
Development of Easybase Release 2.0 has just been started. With its completely new format, it will definitely not risk to infringe copyrights. Its development can also be accelerated. This will allow Release 2.0 to be introduced four or eight months earlier than originally planned, provided it is initiated within a month. The difficulty is that without some preliminary analysis, it is difficult to predict how early Release 2.0 can be introduced. Accelerating will also increase the cost by £100,000 for a four-month early release to £200,000 for an eight-month early release. However DataSoft would recoup about £150,000 of the potential loss on every month of early introduction if they abandon Release 1.1. Both John Boris and Michael Grove also think that Brexoft would be willing to accept an out-of-court settlement for £300,000 in compensation for past copy-right infringements, if DataSoft stopped marketing Release 1.1 by the end of the month. They also think that Brexoft would be willing to accept a licence fee of £625,000 if Release 2.0 is introduced 8 months earlier and £800,000 if it is introduced 4 months earlier.
Michael reminds John that he has at most until the end of the month i.e. 2 weeks to make up his mind wether or not to accept an out of court settlement. Brexoft will not consider such settlement once the court case has been initiated.
There are two major uncertainties in this situation. What are DataSoft's chances of successfully defending a court case? And, how much earlier can Release 2.0 be marketed if DataSoft decides to accelerate its development? Michael estimates that DataSoft has about a 70% chance of winning the case. Prior to a preliminary analysis, the Easybase development leader estimates there is a 60% chance of being able to complete Release 2.0 eight months earlier and a 40% chance of being able to complete it only four months earlier. The development leader suggests that these estimates could be improved by undertaking a preliminary analysis which takes about a week and costs £20,000. This analysis is however not infallible and has a 15% chance to predict a four-months early release when it actually is released eight-months early and a 20% chance to wrongly predict an eight-month early release.
John Boris is very confused. Should he do for a preliminary analysis or not? Should he accelerate or not Release 2.0? Should he go to court or not? As Datasoft's only analyst, John Boris asked you to assist him in his decision making process. John Boris also agrees that his main objective is to maximise the expected profit to be made. In order to advise John Boris, you must answer the following questions. Unless otherwise stated the assumptions are the ones above.
1. Draw a decision tree skeleton. Show all relevant monetary values and probabilities.
2. Build a spreadsheet model based on your decision tree using Precision Tree and solve the tree. You must include a copy of your spreadsheet showing all the formulae and a short report explaining the formulae.
3. What is the highest price that management should pay for the preliminary analysis with its current performance? What is the highest price that management should pay for any preliminary analysis regardless of its performance? Explain.
4. Carry out a complete sensitivity analysis on the probability of being able to complete Release 2.0 eight months earlier (currently 0.6). You may use increments of 0.05. Explain.
5. Carry out a complete sensitivity analysis on both counter-performance probabilities of the preliminary analysis. Explain.
6. Other uncertainties are the future revenue potential for Release 1.1 initially estimated at
£1,500,000 and the chance of winning the case initially estimated at 70%. Revise the decision as the revenue for Release 1.1 varies from £1,000,000 to £2,000,000 and the chance of winning the case varies from 50% to 90%. Explain.
7. Summarise all your recommendations in a short management report (no more than one typed A4 page). Remember that management has no knowledge of decision analysis therefore your report should be as non-technical and as clear as possible.
Probability and Statistics for Engineering and the Sciences
ISBN: 978-1305251809
9th edition
Authors: Jay L. Devore