You are on page 1of 41

Lecture 3

Outline

Variation
Significant Figures
Accuracy
Precision

We need to deal with two types of variations

There are Variations in measured output


on same system (component, device,
sample of material) and
Variations between samples , batches
Conflict between supplier and buyer or
between process engineer and QC engineer
on the source of variations
BE CERTAIN about UNCERTAINTIES in
communication or exchange of data or
representation of data

Data representation
Numerical data
Why is it important to use numerical
representation? Is it always good?
How to represent numbers?
Use the number with proper number of
significant digits with appropriate
dimensions (units)
Caution on dimensionless numbers

Statistics in Engineering
As engineers perform experiments,
they collect data that can be used to
explain relationships better and to
reveal information about the quality
of products and services they
provide.

Frequency Distribution:
Scores for an engineering class are as follows: 58,
95, 80, 75, 68, 97, 60, 85, 75, 88, 90, 78, 62, 83,
73, 70, 70, 85, 65, 75, 53, 62, 56, 72, 79
To better assess the success of the class, we make a
frequency chart:

Now the information can be better analyzed.


For example, 3 students did poorly, and 3
did exceptionally well. We know that 9
students were in the average range of 7079. We can also show this data in a freq.
histogram

Cumulative Frequency
The data can be further organized by calculating
the cumulative frequency (CDF).
The cumulative frequency shows the cumulative
number of students with scores up to and
including those in the given range. Usually we
normalize the data - divide 26.

Extra: Rules for making frequency distribution


chart
A Frequency Distribution is simply asummary of how often each score occursby grouping
scores together.
have to group the scores so that summary makes sense
decidehow many groupsyou want (thefewergroups, theless preciseyour description)
too manygroups, and it'stoo much work, too little summary (usually 10 to 20 groups is best)
take the range(difference between highest and lowest)
divideby number ofintervals(groups) you want
the lowest interval (group) should begin with a number that can be divided evenly by the size
of the interval width; make sure that the lowest and highest scores are included in your table
relationship between these depends on the frequency distribution
Graphs must be clearly titled; all symbols clearly identified
Histogram vertical bar graph; columns touch each other
Frequency distributions come inthree types: symmetrical, positively or negatively skewed
If a distribution issymmetrical, the lower half of the distribution mirrors the upper half
in symmetrical distributions, mean, median and mode will be the same (unless it is bimodal)

Inskewed distributions, median comes between mean and mode


becausemeanwill be dragged down or up by the few scores out on the extreme tail, whilemodewill hang
out where the greatest action is

If the scores are piled up at thelow end, and "tail off" near the high end, it'spositively
skewed
If the scores are piled up at thehigh end, and "tail off" near the low end, then it'snegatively
skewed
PositiveorNegativeby the way the tail points --if tail points to high end, it's positive; if tail
points towards negative end, it's negative

Significant Figures
Accuracy & Precision

Why?
it can often be far too easy to
exaggerate a number's accuracy,
leading to mathematical errors. For
these reasons, significant digits are
important.

Numbers with appropriate digits


Is 1.2340 the same as 1.234?
Multimeters are sold as 4 digits or 4 and half digit
multimeters. More the digits, the cost is more?
Why?
Balance used by subjiwallah is cheaper than that
used by zaveriwallah. Why?
Representing quantity
Numerical value with the correct number of digits
Explicit mention of uncertainty
and of course units (dimensions) where necessary
How many digits to use?

Significant Digits: Number of digits you are sure


of with certainty + the first in doubt
Most significant digit left most non zero digit
(0.52)
If there is no decimal point, the right most nonzero digit is least significant
Population 105 crores or 1050000000
If there is a decimal point, right most digit is
least significant even if it is zero
All digits in between the least and most
significant digits are taken to be significant

Find significant digits in 1234,


1234000,123.4,1,1001,1000,10.10
,0.0001010
1234, 1234000,123.4,1,1001,1000,10.10,0.0001010
all have four significant digits
1.2340(0 is least significant and 1 is most significant,
there are FIVE significant digits)
1.2340 is not the same as 1.234
0.1234(1 is most significant and 4 is least
significant)
0.001234(also has four significant digits)
All digits between and including most and least
significant digits are significant. The least has a
builtin uncertainty

Rounding off Rules


If the digit to be reduced is more than 5, the
immediately more significant digit is increased by one
(1.2346 is rounded off to 1.235)
if the digit is less than five, the immediately more
significant digit is not increased by one(1.2343 is
rounded off to 1.234)
What if the digit to be reduced is 5?
Say 1.2345 or 1.2335,
Increment the last significant digit by one only if it is odd
1.2335 will be rounded off to 1.234
1.2345 will also be rounded to 1.234

Rules! (repeat)
Every approximate number has a specific number of significant
digits (or significant figures).
These are digits in the number which convey actual numerical
information, and are not just written down to show us where
the decimal point is located. Thus
(i) all nonzero digits are significant.
(ii) all zeros which are between nonzero digits are
significant.
(iii) all zeros to the right of the decimal point are
significant if they follow nonzero digits in the
number.
(iv) zeros which are present only to show the
position of the decimal point are not significant.
(v) zeros which can be omitted without affecting
the numerical value are not significant.
(This rule overlaps rule (iv.), but includes so-called leading
zeros sometimes written in the whole number parts of
decimal values.)

Zeros!
The first rule covers most situations. The
tricky cases are situations with digits which
are zeros, because the digit zero has two
roles in decimal numbers.
One role is to indicate the value zero at a
certain position (as in 105 equals one
hundred plus zero tens plus five ones).
The other role is to tell us where the
decimal point should be located (as in a
number like 0.0035).

Types of Zeros
Zero Type #1: Space holding zeros on
numbers less than one. Eg
0.00500,0.03040
They are there to put the decimal point in
its correct location. They DO NOT involve
measurement decisions. Upon writing the
numbers in scientific notation (5.00 x
103and 3.040 x 102), the non-significant
zeros disappear.

Zero Type #2: the zero to the left of the


decimal point on numbers less than one.

When a number like 0.00500 is


written, the very first zero (to the left
of the decimal point) is put there by
convention. Its sole function is to
communicate unambiguously that the
decimal point is a decimal point. If the
number were written like this, .00500,
there is a possibility that the decimal
point might be mistaken for a period.
Many students omit that zero. They

Zero Type #3: trailing zeros in a


whole number eg 200,2000
This is based on the way each number is written. When whole
number are written as above, the zeros, BY DEFINITION, did
not require a measurement decision, thus they are not
significant.
However, it is entirely possible that 200 really does have two
or three significant figures. If it does, it will be written in a
different manner than 200
Typically, scientific notation is used for this purpose. If
200 has two significant figures, then 2.0 x 102is used.
If it has three, then 20.0 x 102is used. If it had four,
then 200.0 is sufficient.
How will you know how many significant figures are in a
number like 200? In all such problem, the context of the
experiment and its measuring devices would tell you how
many significant figures to report to people who read the
report of your work

Zero Type #4: leading zeros in a


whole number.
00250 has two significant figures.
005.00 x 104has three.

Exact Numbers have an infinite


number of significant figures
Exact numbers, such as the number of people in a room, have
an infinite number of significant figures. Exact numbers are
counting up how many of something are present, they are not
measurements made with instruments. Another example of
this are defined numbers, such as 1 foot = 12 inches. There
are exactly 12 inches in one foot. Therefore, if a number is
exact, it DOES NOT affect the accuracy of a calculation nor the
precision of the expression. Some more examples:
There are 100 years in a century.
2 molecules of hydrogen react with 1 molecule of oxygen to
form 2 molecules of water.
There are 500 sheets of paper in one ream.
Interestingly, the speed of light is now a defined quantity. By
definition, the value is 299,792,458 meters per second

Examples
142.56, 3001.378,5.40,

3001.378 has seven significant digits. It has seven digits and all seven are
significant. Five of the seven digits are nonzero digits, and so are significant by
rule (i.). The two zeros are between significant digits, and so are themselves
significant by rule (ii.). You can think of these two digits as being as significant
as the other digits in the number because they indicate the value here, for
instance, has no hundreds (as opposed to one hundred, or two hundred etc.)
and that it also has no tens (as opposed to one ten, or two tens, etc.)
5.40 has three digits and all three are significant. The two nonzero digits are
significant by rule (i.). The zero at the end is also significant by rule (iii.) it is
to the right of the decimal point and follows nonzero digits in the number. This
zero indicates that the number has been measured to two decimal places (or,
its uncertainty of measurement is 0.005, following the ideas described in the
previous slide). If the zero in this number was not the result of measurement
(that the value has 5 ones and 4 tenths, and zero hundredths, as opposed to
one hundredth or two hundredths, etc), then it should not have been written.

Examples containing numbers containing


digits which are not significant
00352.6, 0.00516, 0.05016, 0.050160
00352.6 has four significant digits. Obviously the two zero digits on the left convey no information
its not clear why they were even written down here. This is an example of rule (v.).
0.00516 has three significant digits. The three zero digits here are intended simply to show the
location of the decimal point. This is an example of rule (iv.).
0.05016 has four significant digits. The zero between the 5 and the 1 is significant, by rule (ii.).The
two zeros on the left are not significant by rule (iv.). You might argue that all three zeros should be
considered significant because just as the rightmost zero (between the 5 and the 1) indicates no
thousandths, so, the leftmost zero indicates no ones, and the second zero indicates no tenths. But
this argument misses an important point. Suppose this value was the length of an object measured
in metres: length of this object = 0.05016 m If we simply re-expressed this length in millimetres (we
would need to multiply the number of metres by 1000 to get the equivalent length expressed in
millimetres), we could then write length of this object = 5.016 mm This is still the same physical
measurement of the same physical length but now those two zero digits on the left are not required.
Therefore they could not have been significant. This sort of observation is the source of rules (iv.)
and (v.), indicating when zero digits are not significant.
0.050160 has five significant digits. The two zeros on the left are not significant by rule (iv.), as
discussed in the previous example. The third zero is significant by rule (ii.) it is between two other
significant digits. The fourth zero is significant by rule (iii.) it is to the right of the decimal point and
to the right of a significant digit. The idea is that if this rightmost zero was not significant, it would
not have been written down at all.

Finally!
Significant digits in 0.00?

More rules! (repeat)


When adding or subtracting numbers, the
number of digits to the right of the decimal
point in the result should be the same as the
number of digits to the right of the decimal
point in the number with the fewest digits to
the right of the decimal point
When multiplying or dividing numbers, the
number of significant figures in the result is
the same as the least number of significant
figures in any of the multiplied or divided
terms

Do the following operations and write the results with


the correct number of significant figures:
1) 75m x 4m =
2) 75cm x 4.0cm =
3) 0.750 ft x 4.000 ft =
4) 7500 in. x 0.004 in.=
5) 125m / 25s =
6) 80f t / 16s =
7) 33,333mi / 3h =
8) 3750km / 2.50s =
9) (25m - 16m) / 0.0003s =
Answers: 3x102m2(1 sig. fig.),
3.0x102m2(2 sig. fig.),
3.00ft2(3 sig. figs.),
3.0x101in2(2 sig. fig.),
5.0m/s (2 sig. fig.),
5ft/s (1 sig. fig.)
10,000 mi/h (1 sig. fig.),
1.50x103km/s (3 sig. figs.),
30000 m/s (1 sig. fig.)

Accuracy & Precision


The data has to be good enough for the decisions we are making:
the precision and detail of the data actually get in the way of
making good decisions.

In the cartoon, the extra precision on the left actually makes things worse for our poor analyst
(who is about to be hit by a piano). The analyst has to spend too much time trying to understand
the data and misses the opportunity to take the much-needed action of getting out of the way

Accuracy & Precision


By the accuracy of an approximate number,
we mean the number of significant digits it
has.
By the precision of an approximate number,
we mean the actual position of the rightmost
significant digit. If that digit is to the right of
the decimal point, we state the position as so
many decimal places. If that position is to the
left of the decimal point, it is more common to
state the precision using words like tens,
hundreds, thousands,

Instruments typically provide very precise


readings, but accuracy requires calibration
classic illustration distinguishing the two is to consider a
target or bullseye. Arrows surrounding a bullseye indicate a
high degree of accuracy; arrows very near to each other
(possibly nowhere near the bullseye) indicate a high degree
of precision. To be accurate an arrow must be near the
target; to be precise successive arrows must be near each
other. Consistently hitting the very center of the bullseye
indicates both accuracy and precision.

Examples
321.56 has an accuracy of 5 significant digits and
a precision of 2 decimal places.
3.2156 has an accuracy of 5 significant digits and
a precision of four decimal places.
321560 has an accuracy of 5 significant digits
and a precision of tens (assuming the rightmost
zero is not significant.
0.00000003 has an accuracy of 1 significant digit
and a precision of 8 decimal places.
325,000,000 has an accuracy of 3 significant
digits and a precision of millions.

A number can have high accuracy


(many significant digits) but not very
high precision. Similarly, a number
can be very precise, but not very
accurate. This is why we need two
different terms here you cannot say
accuracy when you mean
precision, or vice versa.

Repeat: Statistical Approach


The field of statistics deals with the
collection, presentation, analysis,
and use of data to
Make decisions
Solve problems
Design products and processes

Descriptive statistics
representation of collected data in the form of
Tables
Charts
Inferential statistics
Generalization from samples
Limited observations to predictions, risk
assessment, reliability
Limitations of inference arise from Sampling
error(Subjectivity in data collection, wrong theory..

Statistical techniques are useful for


describing and understanding variability
By variability, we mean successive
observations of a system or phenomenon
do not produce exactly the same result
Statistics gives us a framework for
describing this variability and for learning
about potential sources of variability.

VARIABILITY and QUALITY


ASSURANCE
Consider the following
Differences among different components manufactured
Differences among different batches of materials
(polymers, ceramics, composites manufactured
Differences between batches of chemicals..
CAS # same. But differences during minor constituents
Pharma impurity profile
Specifying the component is easier than specifying
materials

The differences mentioned previously


are random
Supply chain variations (Raw Material)
lead to variations in products (Paste as
eg.)
Supply chain management
Quality control and quality assurance

Origin of Variations / uncertainty?


Tools operator, ambience and several
unknown variables which we assume to be
invariant and they may not be
How do you choose the material for a linear
scale?
How do you choose a multimeter?
How do you choose a balance?

Every experiment and data set has


some inherent variation that must be
understood before inferences about data
can be made
No two measurements are ever exactly
the same, due to both process and
measurement variability
We must always gather a sample of
several data points in order to make
valid inferences

Virtues of numerical data:


Representing data by numbers helps clarify and
simplify communication
Can show general trend or common denominator
in the data
Danger of numerical data:
Liable to misrepresentation or manipulation
can over simplify/ignore individual differences
AGAIN There are lies, damned lies, and statistics
--Mark Twain