Scribd
Upload
21 views

Lecture 3

This document outlines the topics covered in Lecture 3 of EMCH361. [1] It lists the chapters from the textbook that were discussed, including basic concepts, report writing, data analysis, displacement and velocity measurement, circuits, strain measurement, force and pressure measurement, and temperature measurement. [2] It then covers the concepts of statistical analysis of experimental data, including defining the arithmetic mean and using regression analysis to determine the best estimate of data and fit curves to measured data points. [3] Examples are provided to demonstrate calculating the correlation coefficient to determine the quality of fits from regression analysis.

Uploaded by

Icy45
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
21 views

Lecture 3

This document outlines the topics covered in Lecture 3 of EMCH361. [1] It lists the chapters from the textbook that were discussed, including basic concepts, report writing, data analysis, displacement and velocity measurement, circuits, strain measurement, force and pressure measurement, and temperature measurement. [2] It then covers the concepts of statistical analysis of experimental data, including defining the arithmetic mean and using regression analysis to determine the best estimate of data and fit curves to measured data points. [3] Examples are provided to demonstrate calculating the correlation coefficient to determine the quality of fits from regression analysis.

Uploaded by

Icy45
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 25

EMCH361

Professor Y.J. Chao

Lecture #3
Fall 2016

1
CLASSROOM TOPICS

(attempts to coincide with and support Laboratory topics)

Chapter 1: Basic Concepts


Chapter 12: Report Writing
Chapter 11: Analysis of Experimental Data
Chapter 5: Displacement and Velocity Measurement
Chapter 2: Circuits
Chapter 6: Strain Measurements
Chapter 7: Force, Torque, and Pressure Measurements
Chapter 8: Temperature Measurements

Technical Presentation 2
Statistical Analysis of Experimental Data
• For a given measurement of a physical quantity x
• Scattered data
• What is the “best” value X

of x ?

One variable X

3
Statistical Analysis of Experimental Data

The arithmetic mean X


is defined as :

Xm

where N is the total number of


data and xi is the data.
(3.3)

It is an average !!!
4
The deviation of each reading - di
X
(3.4)
d2
xm

d2=x2-xm

5
Regression analysis:
Minimization of a function to achieve the best estimate of data

So, the mean value is the best estimate from regression analysis
6
--- one variable !
Least square for two variables (x,y)

• Considering the linear case

• Two unknowns a and b

Measured data (xi, yi)

From calculus -
7
Equation for the straight line 8
How good is the fit?

9
Least square for higher order polynomials
(for example)

S is the function to be
minimized
(xi, yi) are measured data

10
Least square for higher order polynomials
(for example)

Three equations for three unknowns a, b and c.


Done!

11
Regression analysis, curve fitting (using least
square method)

The best fitted


curve (to the
measured data)

12
Regression analysis, curve fitting (using least square
method)

• If both x and y have uncertainty

Regression of y on x
Regression of x on y
Uncertainty in y dominates
Uncertainty in x dominates

Some sort of averaging must be used.


13
How good is the fit after a regression analysis?
The correlation coefficient

• 0<r<1 in general
• r =1, perfect fit
• Good fit : r as close to unity as possible 14
Regression analysis, curve fitting (using least square
method) – Example

Ans: use

15
Regression analysis, curve fitting (using least square
method) – Example (continued)

16
Regression analysis, curve fitting (using least square
method) – Example (continued from 3.19)

• Calculate the correlation coefficient “r”


• Ans: using

• => r=0.9178
• (good!)

17
Rule of thumb in regression analysis (or curve fit)

• Always plot the data to obtain a visual observation (see


the trend and see any unusual measured data) before
performing any analysis
• Consider taking out those “way out of line” data (if one
knows the functional form)
• Run a regression analysis
• Plot both the data and the fitted curve and see the
correlation between the two
• Never accept a least-squares analysis based only on
calculations

18
19
Example

20
Excel program gives (a straight line fit –with
least square analysis)

Straight line in log-


log scale

21
Choices of x-y
graph format
(using a
computer)

22
Choices of x-y graph format (using a computer)

Symbols: Test data


Lines: from equation

23
Plot directly - sequence
24
EMCH361

END of
Lecture #3

25

You might also like