Question

3)      A project manager at System Design Corporation has collected the following data on the size of software programs and the length of programming time. The company is bidding on a new system that is estimated to consist of 300,000 lines of code. Use a scatter plot to find the correlation function of coding time to the program size and estimate the number of days it would take to code the system.

Module	Size in 1000 lines of code (KLOC)	Number of days to code
1	150	72
2	158	66
3	148	65
4	120	60
5	178	75
6	170	68
7	152	68
8	138	65
9	200	80
10	195	75
11	189	65
12	173	68
13	159	62
14	163	71
15	150	65
16	140	65
17	206	74
18	144	64
19	157	70
20	183	76
21	195	75
22	190	77
23	185	73
24	152	58
25	178	70

Answer 1

Let x = Size in 1000 lines of code (KLOC) and y = Number of days to code.

Sample correlation, r = Corr(x,y) = 0.772095 (please round this number to required number of decimal places).

Now, for 300,000 lines of code, that is, x = 300, we get estimated number of days to code = = 36.5435 + (0.194923 * 300) = 95.0204 = 95 days (approx.).



From the scatterplot, we observe that there exists a strong, uphill (positive) linear relationship between x = Size in 1000 lines of code (KLOC) and y = Number of days to code.