Hw4 – Computer system simulation Fall 2023

– HW3 can be done as a group of 2 students, maximum. Individual submissions will get 20%
bonusf= max [0.2, (75)/100] = max [0.2, 0.75] =0.75 , If individual

Take n1=30f=30*0.75=23, n2=70+30f =70+30*0.75=93 (round)

Q1. Following is a partial Chi-square table to check the goodness-of-fit for a normal distribution (mean=250,
standard dev. = 40), Assume that the number of data points is N=125. Complete the Chi-square test.
Note:Sum of the Oi s may not be equal to 125 in this question.

i Interval i Oi Ei = p*i = (Oi-Ei)^2/Ei

(1/10) *125
1 "-infinity" 198.74 13 12.5 (13-12.5)^2/12.5=
0.02
2 198.74 216.34 14 12.5 (14-12.5)^2/12.5=
0.18
3 216.34 229.02 17 12.5 (17-12.5)^2/12.5=
1.62
4 229.02 239.87 18 12.5 (18-12.5)^2/12.5=
2.42
5 239.87 250 13 12.5 (13-12.5)^2/12.5=
0.02
6 250 260.13 10 12.5 (10-12.5)^2/12.5=
0.5
7 260.13 270.98 16 12.5 (16-12.5)^2/12.5=
0.98
8 270.98 283.66 17 12.5 (17-12.5)^2/12.5=
1.62
9 283.66 301.26 12 12.5 (12-12.5)^2/12.5=
0.02
10 301.26 "+infinity" 20 12.5 (20-12.5)^2/12.5=
4.5
sum 125 11.88
dof = 10-2-1=7
2
𝑥0.05,7 = 14.07 > 𝑥0 = 11.88

Null Hypothesis Ho is accepted at the 0.05 level of significance. Good fit.

Q2. Using Data set 1 about the number of calls to an agent in a call centre between 9:00 AM and 11:00 AM
taken in the last 100 days (from data day 23 to day 93).

- Draw a histogram of your data and comment on if it looks like to be close to Poisson distribution or
discrete uniform distribution.
 The histogram doesn’t look like normal nor possion distributions

- Assume it follows Poisson distribution, calculate lambda parameter

500
𝑋̅ = =5
100
𝜆̅ = 𝑋̅ = 5

- Apply Chi-square test to check the goodness-of-fit using the significance level = 0.05.

from data day 23 to day 93:

𝟕𝟎
√𝟕𝟎 = 𝟖. 𝟑𝟔 ≈ 𝟖 𝒕𝒐 𝟓
= 𝟏𝟒
∑𝑘𝑗=1 𝑓𝑗 𝑋𝑗 500
𝑋̿ = = =5
𝑛 100
𝐸𝑖 = 𝑛 ∗ 𝑝(𝑥𝑖) = ((𝐸𝑋𝑃(−5) ∗ (5^𝑥𝑖)/𝑥𝑖! ∗ 100

(Oi-
xi*Oi xi Oi Ei
Ei)^2/Ei
8 1 8 3.368973
18 11.79141 3.269044
20 2 10 8.422434
45 3 15 14.03739 0.066011
60 4 15 17.54674 0.369634
50 5 10 17.54674 3.245802
72 6 12 14.62228 0.470266
49 7 7 10.44449 1.135957
112 8 14 6.527804 8.553216
54 9 6 3.626558
9 5.439837 2.32999
30 10 3 1.813279
500 100 19.43992

Number of intervals = 8
Degree of freedom = k – s – 1 = 8 – 1 – 1 = 6

𝑥02 = 19.43992 > 𝑥0.05,6

2
= 12.59

Null Hypothesis Ho is rejected at the 0.05 level of significance. Not a good fit.

Q3. Using Data set 2 about the process times taken for the last 100 parts (from data part 23 to part 93).

- Draw a histogram of your data and comment on if it looks like to be close to Normal distribution

The histogram looks like normal distribution.

- Assume it follows normal distribution, calculate its parameters and draw Quantile-Quantile plot and
comment about the plot.
∑50
𝑥=1 𝑥
𝜇= = 4.0664
50
2
∑50
𝑥=1(𝑋 − 𝑋 )
̿ ∑50
𝑖=1(𝑋 𝑖 − 4.0664)
2
𝜎2 = = = 0.574564
𝑛−1 50 − 1
𝜎 = 0.758

Quantile-Quantile plot R² = 0.992

7

0
2 2.5 3 3.5 4 4.5 5 5.5 6

The points in the middle fall approximately in a straight line, the variance at the extremes (as shown
from point [2,3) and (4.8,5.5)) is higher that the middle

Draw scatter diagram (XY plot in excel) between two successive data (X axis is Xi is, Y axis is Xi+1) and
comment about the dependency of successive data points.

scatter diagram R² = 0.992

7
6
5
4
3
2
1
0
2 2.5 3 3.5 4 4.5 5 5.5 6

The correlation coefficient = 0.992, which is close to 1. it indicates that there is a strong positive
correlation, as we increase one value, the other value also increases.

- Calculate the correlation between two successive data (Xi is one column, Xi+1 is the other column of
data) and comment about the dependency of successive data points. (You can use the correlation
function of excel, or formulae given our classes.)
Correlation coefficient = 0.992

- Apply Chi-square test to check the goodness-of-fit using the significance level = 0.05. What is the p-
value of your test?
 𝟕𝟎
- √𝟕𝟎 = 𝟖. 𝟑𝟔 ≈ 𝟖 𝒕𝒐 𝟓
= 𝟏𝟒
Ei = np

(Oi-
interval start interval end Oi Ei Ei)^2/Ei
0 2.640758445 4 4.6 0.078261
2.640758445 3.136691685 10 4.6 6.33913
3.136691685 3.506354088 0 4.6 4.6
3.506354088 3.774327086 8 4.6 2.513043
3.774327086 4.009345445 0 4.6 4.6
4.009345445 4.238879092 5 4.6 0.034783
4.238879092 4.485865617 0 4.6 4.6
4.485865617 4.789657262 9 4.6 4.208696
4.789657262 5.313199049 10 4.6 6.33913
5.313199049 infinity 0 4.6 4.6
46 46 37.91304

dof = 10-1-1=8

𝑥02 = 37.91304 > 𝑥0.05,8

2
= 15.51

Null Hypothesis Ho is rejected at the 0.05 level of significance. Not a good fit. p value is closest to 0, because
we reject.

Q4. Fill in the gap

- Time to failure for a computer network which fails when a computer and two backup computers fail
and each has a time to failure that is exponentially, is expected to follow Erlang distribution.
- Process time in a machining centre depends on several random factors that affect it in positive or
negative fashion is expected to follows a normal distribution.
- Manual assembly processes usually follows Weibull distribution
- Time between failures in a milling machine that fails mainly due to mechanical wear is expected to
follow normal distribution
- Life time of a microchip (single component) is expected to follow exponential distribution.
- Lead time of orders in an inventory system is expected to follow gamma distribution.
- If for a process there are clear min and maximum values, and most of the process time concentrates
around a particular value than we can expect the process time to follows triangular distribution
- Chi-square test should not be used if does not meet the assumptions of independence and normality
- Shape-wise two very flexible distributions are normal distribution and the t-distribution.

To be submitted:

- All of your answers as a word file (embed results, calculations, plots etc. from your excel file). This
word file will be graded, not excel file.
- Also submit your excel file (We might look at it, what will be graded is the word file. So you must
submit a word file)
Data Set 1

Number Number
Day of calls Day of calls
1 10 51 1
2 3 52 3
3 8 53 9
4 10 54 6
5 10 55 4
6 6 56 6
7 5 57 3
8 3 58 5
9 9 59 2
10 5 60 6
11 4 61 8
12 8 62 3
13 2 63 2
14 8 64 8
15 6 65 6
16 4 66 3
17 5 67 5
18 2 68 1
19 8 69 1
20 4 70 8
21 7 71 2
22 8 72 4
23 6 73 2
24 8 74 6
25 3 75 8
26 1 76 9
27 6 77 3
28 1 78 6
29 4 79 4
30 4 80 1
31 9 81 3
32 7 82 5
33 6 83 5
34 4 84 1
35 2 85 5
36 5 86 8
37 7 87 4
38 4 88 3
39 2 89 9
40 8 90 3
41 2 91 3
42 4 92 7
43 4 93 4
44 1 94 2
45 8 95 7
46 6 96 7
47 8 97 3
48 5 98 4
49 3 99 9
50 7 100 3
Data set 2

Process times Process

Part No (min) Part No times (min)
1 3.64 51 2.6
2 4.68 52 4.68
3 3.64 53 4.16
4 3.12 54 3.64
5 4.16 55 3.64
6 3.12 56 5.2
7 3.64 57 4.68
8 5.2 58 3.64
9 4.16 59 4.16
10 4.16 60 3.64
11 4.68 61 4.68
12 4.68 62 4.68
13 5.2 63 4.16
14 3.64 64 4.16
15 4.16 65 5.2
16 3.64 66 3.64
17 4.68 67 2.6
18 2.6 68 5.2
19 3.12 69 3.12
20 3.12 70 3.12
21 5.2 71 2.6
22 4.68 72 4.68
23 5.2 73 3.64
24 2.6 74 5.2
25 3.12 75 2.6
26 4.68 76 3.64
27 5.2 77 3.64
28 4.68 78 4.68
29 3.64 79 5.2
30 3.64 80 4.68
31 5.2 81 3.64
32 2.6 82 2.6
33 3.12 83 5.2
34 3.64 84 3.64
35 4.16 85 3.12
36 2.6 86 3.64
37 4.68 87 3.64
38 4.68 88 3.12
39 4.68 89 4.68
40 3.12 90 5.2
41 5.2 91 4.16
42 4.16 92 2.6
43 3.12 93 4.68
44 5.2 94 4.68
45 3.12 95 4.68
46 3.12 96 4.16
47 5.2 97 5.2
48 5.2 98 3.64
49 3.64 99 4.68
50 5.2 100 5.2