Assume that you are a sales manager for a telecommunication company and that your company has a mobile phone operation. Due to the introduction of a special subscription plan by your competitor, you have lost some customers to them within the first several days. In order to prevent further loss, your manager would like you to take a close look at the following data set sampled from those customers who responded to a survey. In the survey, the customers were asked if they would remain with the current subscription plan or switch to the competitor’s.
Customer Average Average Total Decision
No. Monthly Payment Duration of Calls Calling Time
1 215.16 6.84 41.63 Stay
2 320.32 14.76 97.25 Switch
3 218.24 3.96 124.88 Stay
4 352.00 8.04 56.00 Undecided
5 259.16 11.88 82.50 Stay
6 462.44 2.64 162.38 Undecided
7 220.66 2.16 69.75 Switch
8 214.72 8.04 96.88 Switch
9 373.78 9.24 122.50 Undecided
10 409.86 17.28 180.50 Switch
11 378.62 18.60 83.50 Undecided
12 195.36 5.88 138.88 Switch
13 291.94 9.00 111.13 Stay
14 342.10 12.00 96.38 Stay
15 317.68 10.68 126.38 Switch
16 197.54 4.4 147.3 Stay
17 404.43 17.4 171.7 Switch
18 264.11 5.7 107.1 Switch
19 430.44 3.14 100.74 Undecided
20 273.43 8.7 98.7 Undecided
a) Assume that k = 5, using the k-NN algorithm, what do you expect the decision of a customer, who has an average monthly payment of 293.26, an average duration of calls of 6.96 and a total calling time of 110.25, to be?
b) Assume again that k = 5 and ignoring the "decision" of the customers. Using the k-NN algorithm, what do you expect the average duration of calls of a customer to be, given that his average monthly payment of 271.48 and a total calling time of 114.00?
c) If you are free to choose the value of k, what will your choice be? Why?
[返回] 计算机科学论坛 →
计算机理论与工程 → 『 人工智能 :: 机器学习|数据挖掘|进化计算 』 → 各位懂数据挖掘的大侠们。。都过来帮帮忙啦。。。
(订阅本版)
2008/5/5 12:23:00
新浪微博
各位懂数据挖掘的大侠们。。都过来帮帮忙啦。。。
