You are on page 1of 11

Design of

Experiments
Case Study analysis
Kishore.P.Chakravarthy
1MS12MIA04
M.tech 2nd SEM
PROBLEM STATEMENT-
Do you risk a ticket by parking where you should not or forgetting how much time you have left
on the parking meter? Do the fines associated with various parking infarctions vary depending on the
city in which you receive the ticket? To look at this issue, the fines imposed for overtime parking,
parking in a red zone and parking next to a fire hydrant were recorded for 13 cities in southern
California.
CITY OVERTIME
PARKING($)
RED
ZONE($)
FIRE
HYDRANT($)
LONG BEACH 17 30 30
BAKERSFIELD 17 33 33
ORANGE 22 32 30
SAN BERNARDINO 20 78 30
RIVERSIDE 21 30 30
SAN LUIS OBIPSO 8 75 20
BEVERLY HILLS 23 30 38
PALM SPRINGS 22 46 28
LAGUNA BEACH 22 32 22
DEL MAR 25 55 40
LOS ANGELES 20 30 55
SAN DIEGO 35 60 60
NEWPORT BEACH 32 30 42

1. Identify the design used for the data collection in this case study
2. Analyze the data using appropriate analysis. What can u say about the variation among the cities
in this study and among fines for three types of violations/ can turkeys test be used in further
delineating any significant differences you may find?
3. Provide interaction, main effects plot, check for the validity of the model for the above data.
OBJECTIVE-
To find whether the fines associated with various parking infractions vary depending on the city
in which ticket is received?
Design for the case study.
To obtain main plots, interaction plot, and check for the validity of the model.
Turkeys test to de-lineate problem.
METHODOLOGY-
Full Factorial Design (Multi factors) using MINITAB software.
-Factor A- Cities & Levels- 13 cities
-Factor B- Type of fine & Levels- overtime parking, red zone, and fire hydrant (3).
MODEL-
The 2k designs are a major set of building blocks for many experimental designs. These designs are
usually referred to as screening designs.
In general for 2k factorials the effect of each factor and interaction is:
Effect = (1/2(k-1) n) /[contrast of the totals]
Variance (Effect) = 2 / 2(k-2) n
SS (Effect) = (contrast of totals)
2
/ 2kn

DESIGN
Std Order Run Order Pt Type Blocks City Fine Type Parking Fines
($)
29 1 1 1 Del Mar Red Zone($) 40
18 2 1 1 San Luis
Obispo
Fire
hydrant($)
75
39 3 1 1 Newport
Beach
Fire
hydrant($)
30
8 4 1 1 Orange Red Zone($) 30
1 5 1 1 Long beach Overtime
Parking($)
17
24 6 1 1 Palm Springs Fire
hydrant($)
46
38 7 1 1 Newport
Beach
Red Zone($) 42
30 8 1 1 Del Mar Fire
hydrant($)
55
2 9 1 1 Long beach Red Zone($) 30
21 10 1 1 Beverly Hills Fire
hydrant($)
30
33 11 1 1 Los Angeles Fire
hydrant($)
30
31 12 1 1 Los Angeles Overtime
Parking($)
20
11 13 1 1 San
Bernardino
Red Zone($) 30
36 14 1 1 San Diego Fire
hydrant($)
60
25 15 1 1 Laguna
Beach
Overtime
Parking($)
22
9 16 1 1 Orange Fire
hydrant($)
32
4 17 1 1 Bakersfield Overtime
Parking($)
17
32 18 1 1 Los Angeles Red Zone($) 55
17 19 1 1 San Luis
Obispo
Red Zone($) 20
28 20 1 1 Del Mar Overtime
Parking($)
25
13 21 1 1 Riverside Overtime
Parking($)
21
35 22 1 1 San Diego Red Zone($) 60
3 23 1 1 Long beach Fire
hydrant($)
30
16 24 1 1 San Luis
Obispo
Overtime
Parking($)
8
5 25 1 1 Bakersfield Red Zone($) 33
27 26 1 1 Laguna Beach Fire
hydrant($)
32
6 27 1 1 Bakersfield Fire
hydrant($)
33
37 28 1 1 Newport
Beach
Overtime
Parking($)
32
19 29 1 1 Beverly Hills Overtime
Parking($)
23
20 30 1 1 Beverly Hills Red Zone($) 38
12 31 1 1 San
Bernardino
Fire
hydrant($)
78
26 32 1 1 Laguna Beach Red Zone($) 22
10 33 1 1 San
Bernardino
Overtime
Parking($)
20
34 34 1 1 San Diego Overtime
Parking($)
35
7 35 1 1 Orange Overtime
Parking($)
22
14 36 1 1 Riverside Red Zone($) 30
23 37 1 1 Palm Springs Red Zone($) 28
22 38 1 1 Palm Springs Overtime
Parking($)
22
15 39 1 1 Riverside Fire
hydrant($)
30

ANALYSIS OF VARIANCE FOR PARKING FINES ($)
Analysis of variance (ANOVA) is a collection of statistical models used to analyze the differences
between group means and their associated procedures (such as "variation" among and between
groups).
Source Dof Sum of squares Mean squares F value
City 12 2122.10 176.84 -
Fine type 2 3015.70 1507.84 -
City*fine type 24 3943.59 164.32 -
Error 0 - -
Total 38 9081.44

By analyzing from ANOVA we can conclude that city has less significance compared to the fine type and
interaction between city and fine type, i.e., 24.33% of the total value.
MAIN PLOTS
City Least Squares Means
Long beach 25.667
Bakersfield 27.667
Orange 28.000
San Bernardino 42.667
Riverside 27.000
San Louis Obispo 34.333
Beverly hills 30.333
Palm springs 32.000
Laguna beach 25.333
Del mar 40.000
Los Angeles 35.000
San Diego 51.667
Newport beach 34.667




FINE TYPE LEAST SQUARE MEANS
Overtime Parking($) 21.84
Fire hydrant($) 35.23
Red Zone($) 43.15


By analyzing from the main plots we can interpret that cities such as SAN BERNARDINO, LAGUNA
BEACH, and SAN DIEGO are causing little significance. In general we can interpret that they are not
much significant compared to the type of fines. We can infer it from the graph as shown, slope of
fine type is more than that of the cities.
INTERACTION PLOTS
City*Fine Type LEAST SQUARE MEANS
San Bernardino *Fire hydrant($) 78
San Luis Obispo* Fire hydrant($) 75
San Diego * Red Zone($) 60
San Diego * Fire hydrant($) 60
Del Mar * Fire hydrant($) 55
Los Angeles * Red Zone($) 55



As mentioned from the graph we can conclude that cites along with parking fine types have more
significance compared to the main effects. Cities such as SAN BERNARDINO, NEPORT BEACH, SAN
DIEGO, LOS ANGELES interacts more with the type of parking fines. From this we can infer that this
system also follows PARETO RULE.

TURKEYS TEST-
It is a single-step multiple comparison procedure. It is used in combination with an ANOVA to find
means that are significantly different from each other.
Turkeys test compares the means of every treatment to the means of every other treatment; that is, it
applies simultaneously to the set of all pair wise comparisons and identifies any difference between two
means that is greater than the expected standard error.




TURKEYS TEST FOR FINE TYPE
Source Dof Sum of
squares
Mean
squares
F value P value
Fine type 2 3015.70 1507.84 8.95 0.001
Error 36 6066 168
Total 38 9081.44


TURKEYS TEST FOR CITY
Source Dof Sum of
squares
Mean
squares
F value P value
CITY 2 2122 177 0.66 0.772
Error 36 6959 268
Total 38 9081

To confirm that city has less significance, turkeys test is conducted and from the above table we can
infer that city has less significance. (If p value is high then it is less significant)






TO CHECK THE ADEQUACY OF THE MODEL-

The normal probability plot is a graphical technique for normality testing: assessing whether or not a
data set is approximately normally distributed. The data are plotted against a theoretical normal
distribution in such a way that the points should form an approximate straight line. Departures from
this straight line indicate departures from normality.

In the problem we can infer that residuals are following the normal distribution with the mean 0.
Hence its a good fit.
NORMAL CURVE
CONCLUSIONS
Based on the results we can conclude that main effects have less significance.
Interaction effects have more significance.
By TURKEY TEST we can confirm that cities have less effect.
Based on normal probability plot the model is adequate.

You might also like