代做CSCU9T6、代做c/c++，Java编程语言、代做Python设计

日期：2020-03-02 10:21

Ivor Buquetlowd

Managing Director

World of Bargains

Bargain House

MK5 6LP

Dear CSCU9T6 Student,

Let me introduce myself, I am Ivor Buquetlowd, the owner of a chain of over 100

shops in the UK and I would like you to help me run them more efficiently. The shops

are all similar, but they make amounts of revenue varying from ￡1 million to over ￡5

million per year. I want to understand why some shops are doing better than others. I

also want a computer program that lets me enter the details of a shop and gives me a

prediction of how much money that shop should make.

I have data describing the following aspects of each store:

? Town

? Store ID

? Manager name

? Staff numbers

? Floor Space and Window Space

? Car park (yes or no)

? Demographic score

? Location (Shopping Centre, High Street, Retail Park)

? 40 min, 30 min, 20 min and 10 min population size

? Store age

? Clearance space

? Competition number (how many competing stores are near ours)

? Competition score (from how good the competing stores are)

? Revenue from last year

I have asked my IT department to make the data for my stores available to you. You

can download it from Canvas.

Please remember that this data is commercially sensitive and you should not show it

to anybody or discuss your findings with anybody except Dr Jingpeng Li.

I would appreciate it if you could use the Weka software to build a predictor capable

of estimating the revenue that a store should generate based on its values for the

variables listed in the data. Remember, it costs a lot of money to collect each value, so

the fewer variables you use, the better. The population data is particularly expensive

to collect, so I would be grateful if you could tell me whether or not I should continue

to use it. However, the system will only be of use to me if the average size of the error

made on predictions is less than half a million pounds.

I would particularly like to know whether a multi-layer perceptron or a decision tree

could help, as my IT director has some experience of these techniques.

Please send me a report detailing the following:

? An introduction to the task and the techniques you used. Please explain why

data mining is suitable for this task;

? Which variables you chose to investigate, and why you chose them;

? A detailed description of the process you followed and the results that you

found. This section should contain a lot of detail. Make sure you include the

following:

o What pre-processing you performed on the data and what problems, if

any, you found and fixed;

o A technical description of the two techniques I’ve asked you to

investigate – how they learn, how they represent the solution and how

they make predictions. Please explain them carefully so that I can

understand the concept and the maths behind it;

o The process you used to arrive at the solution – how did you test it and

how did you establish that it is the best possible solution from the data

you had? Include and explain accuracy measures for each technique

and each subset of variables that you tried;

? A full analysis of the results gained by the best model you could find – how

accurate is it? Where does it make errors?

? A recommendation for which variables I should not pay to collect in future, or

in other words, which variables are sufficient for this task? Describe how you

came to this conclusion;

? A recommendation of which technique I should use and why.

The report is all I need you to send me. If I like the results, I can use Weka to

reproduce them myself. Please make sure your report contains the detail to allow me

to do so.

Good luck and thanks for your hard work.

Best regards,

Ivor Buquetlowd

M.D World of Bargains