# EPH7112 DATA ANALYSIS

Document Sample

```					EPH7112 Research Methodology
CONTENTS

 Planning  for Results
 Generating Data/Results
 Reporting Results
- Presenting Results
- Analyzing Results
 Organizing Results
PLANNING FOR RESULTS

 Requires           Micro-Level Planning : K-
Chart
 Results Layer
Method 1

Results:
Performance
Acceleration        Speed              Fuel
Parameters                                          consumption

Design        Pressure      Tire       Tire Width         Tire     Tire Width
Parameters                 Height                       Pressure
GENERATING DATA/RESULTS
 Again,use Micro-level planning:
K-Chart
 Methodology Layers

Theory         Experiment    Simulation      Survey

Lab. Tests             Field Tests

Lab Prototype   Eng. Prototype        Commercial Prototype

Measurement     Measurement          Measurement
Techn.1         Techn.2              Techn.3
TYPES OF DATA

 Discrete
 Continuous
DISCRETE DATA

 Data which can take only discrete
values: Yes or No, Frequency of
occurrences, Likert scale
 Typically questionnaire based data
 Familiar in social sciences
 Typical analysis:
 Cross-tabulations
 Non-parametric tests
CONTINUOUS DATA
 Data which can take any values
 Typically measured parameters such as
temperature, pressure, weight, height,
frequency spectrum
 Familiar in sciences and engineering fields
 Types of analysis:
- Descriptive analysis
- Relationship analysis
- Comparative analysis
- Optimization analysis
DESCRIPTIVE ANALYSIS
 Statistics of the data
 Typical parameters:
- Max,Min
- Mean
- Standard deviation, Variance
 Mean is mostly used for science and
engineering
 Although seldomly performed, it’s
important to know your data distribution
RELATIONSHIP ANALYSIS

 Also  known as Cause and Effect
analysis
 Normally presented in the form of a
graph of Y versus X
 Y is Effect or Performance Parameter
(PP), X is Cause or Design Parameter
(DP)
 Analyze the Trend and Reason
SAMPLE RELATIONSHIP
ANALYSIS
Y              Y

X              X

Y              Y

X              X
COMPARATIVE ANALYSIS

Y                                     P1
P2

P3

X

Y versus X at various P

•   P is the different environments/setups
•   Be careful when P is another DP
OPTIMIZATION ANALYSIS

 Important   to identify the optimal
conditions
 Cases when:
1. One DP affecting two PPs in
contrasting manner
2. One PP is affected by two DPs in
contrasting manner
OPTIMIZATION CURVES 1
Laser                                       Laser
Power,P      Operating range                Stability,S

Smin

Pmin

Imax
Imin
Laser Current,I

Identify the operation range of the PPs
Important to determine the optimum operating point as well as
the operating range of the DP
OPTIMIZATION CURVES 2
Inflation Rate    Increment of wages
from the existing
Increment of wages
of public servants
Inflation rise

Inflation                                   Goods Productivity
compensated

Required increment of productivity to compensate
the inflation rise due to new higher wages

•    It is easier to use the DPs in the same units
•    Can also have two different units
•    DPs are mutually independent
•    Used to determine ways to offset each other’s effects
REPORTING RESULTS:
Results Report Template

1.   Title of project,
2.   Title of result
3.   Graph presentation,
4.   Setup parameters,
5.   Method of measurement,
6.   Analysis of Trend,
7.   Analysis of Reason,
8.   Comparative Analysis (critical review)
9.   Statement of Achievement of Objective
SAMPLE RESULT REPORT
1. Title of Project : Analysis of Impairment Factors in Fiber Optic Transmission
2. Result Title : Q factor vs PTx
3. Index: R3: PP2-DP1
4. Graph
Low Transmit Power

30

25

Q Factor   20

15                                                      Q factor

10       Stage 1

5                                Stage 2
Stage 3
0
0   -1.0   -2.0    -3.0      -4.0   -5.0   -6.0    -7.0       -8.0   -9.0
Transm it Pow er

5. System setup parameter
Transmission rate = 2.5G, Dispersion = 16.75 ps/nm.km, Attenuation coefficient = 0.2dB/km,
Fibre length = 50km

6. Method of measurement
Simulation- Optisys V.4. Taken after the receiver in electrical domain. Q is calculated value
based on eye pattern
RESULTS REPORT (CONTINUE)
7. Analysis of Trend
The reduction of launching power to -7dBm would give an almost similar effect onto Q
factor as the attenuation coefficient of 0.34dB/km.
Q factor reduces exponentially with launching power reduction. Can be represented by
the exponential equations during 3 stages: Stage 1 :0 to -3.0dBm : y = 33.19e-0.21x,
Stage 2: -3.0 to -6.0dBm : y = 17.562e-0.23x, Stage 3 : -6.0 to -9.0dBm : y = 8.863e-0.23x
Q factor decreases with the steepest curve line for the reduction of launching power from 0
to 3.0dBm. At -7dBm launching power, the Q factor obtained from the simulation result is
5.67864.

8. Analysis of Reason/Discussion
As power reduces, the difference between signal and noise becomes smaller, thus closing
the eye, therefore lower Q. At very low power, signal-independent noise becomes
dominant thus the constant Q value (exponential curve). For higher data transmission rate,
higher launching power would be required but need to watch out for the non linear effect
due to higher power. For low data transmission rate, we can use a lower power laser for
cost effective solution.

9. Comparative analysis
Nothing new

10. Achievement of Objective
Objective 1 to analyze the impairment factor in fiber optic transmission
RESULTS REPORT (CONTINUE)

 May combine    multiple curves on the
same graph if the curves have the same
PP and DP
 The multiple curves should represent
different scenarios/methods
 Combined graphs are good for
comparative analysis, but can lead to
confusion
RESULTS ORGANIZATION TABLE

CODE NUMBER          RESULT TITLE
R1: PP1-DP1    The effect of fiber distance
on BER
R2: PP1-DP2    Optimization of the effects
of Launch Power on SBS
and BER
R3: PP2-DP1    The effect of fiber distance
on SBS threshold

RN: PPY-DPX
PLANNING FOR PAPER WRITING

 Each individual  result or their
combinations may be used for paper
writing
 Results with novelties can be targeted
for high quality journals
 The number of publication can be
maximized
PLANNING FOR THESIS WRITING

 Thesis chapters for results can be
easily planned
 Multiple results chapters can be
generated based on the results
groupings
 Groupings can be based on PPs, or
Methods or Sub-Systems
CONCLUSION

 Results are the most important part of
Research
 Engineering and sciences are
normally dealing with Continuous
data
 Four main analyses: Descriptive,
Relationship, Comparative,
Optimization
 Very important to organize results for
various purposes

```
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
 views: 4 posted: 11/7/2012 language: English pages: 22
How are you planning on using Docstoc?