STAT 111: Introduction To Statistics and Probability: Lecture 2: Data Reduction

STAT 111: Introduction to Statistics
and Probability
Lecture 2: Data Reduction
Ridwan R.D
Department of Statistics and Actuarial Science

University of Ghana
1
Outline
 Populations and samples
 Data types: qualitative versus quantitative
 Data displays: pie chart, bar chart, stem-and-leaf

plot, dot plot, cumulative frequency diagram,
histogram
 Data numerical summaries: sample mean, sample

median, sample variance and sample moments
Populations and samples
Sample is a subset taken from a larger population.
• Example: all the students in this lecture theatre form a

sample taken from the population of all students in UG.
Elements in the sample are called sampling units.
• Example above: students are the sampling units.

Populations and samples
• Random sample is a collection of sampling units chosen

entirely at random. We call this fair sampling.
• If selected unfairly, then it is called biased sample.
Example: in an opinion poll a predetermined proportion is

selected from each social group.
Data types
• Variates are quantities (e.g., height of students) or items
observed (e.g., colour of shoes of students) which are
associated to the sampling units (e.g., the students):
 Qualitative (descriptive, non-numerical)
• Nominal - named, unordered categories
• Ordinal - ordered categories

Data types
 Quantitative (numerical)
• Discrete - distinct values (usually integer values)
• Continuous - any real value in a range
Data consists of a collection of observations (e.g., heights of
students in this lecture theatre).

Qualitative variates
• A variate is qualitative if there is no meaningful way to

perform arithmetic operations with it.
Examples:
• collect autumn leaves; the variate can be the species of

tree from which they come
• observe vehicles passing a census point and record their

type (car, bus, bicycle); the variate here is the vehicle type
• Variates which appear numerical can still be qualitative.
• Example: look at the next 20 buses to go past and record

the numbers on the front. 73 is bigger than 10 as a
number, but a Number 73 bus is not \greater" than a
Number 10 bus - this is simply a characterization of the
buses, the bus number is a nominal (unordered)
qualitative variate
• All previous examples are unordered, also called
nominal, variates. It is possible to have ordered, also
called ordinal, qualitative variates:
• feedback forms have a 5-point scale from ”Very poor" to

“Very good”
• clothes sizes: S, M, L, XL
Quantitative variates
• A quantitative variate is a numerical variate for which a
quantity such as the difference between two observations
can be defined.
• Examples:
• number of passengers in vehicles passing a checkpoint
• heights of students in a class
• volume of noise made by aircrafts taking o
• number of students in lecture theatres The first and last of
these examples are discrete, the second and third are
continuous.
Data displays
A good visual display aids understanding and can highlight
features which may be worth exploring.
• Displays should be appropriate, have impact and be

accurate.
• Displays should be uncluttered and straightforward to
grasp.
• A graphical summary of the data is a very useful starting
point.
Pie and bar charts
• Both useful for categorical data with a few number of

categories.
• In a pie chart, the areas of a segment represents the

category value.
• In a bar chart, the height of a bar represents the value.
• Generally bar charts are better for display. Relative lengths

are easier to judge than relative areas.
Good and bad examples: Bar chart
eg
Steps in constructing a bar chart
1. Draw a horizontal line with category names at regular

intervals underneath.
2. Draw a vertical axis scaled to the largest frequency

value.
3. Above each category draw a rectangle with height at the

category's frequency.
Steps in constructing a bar chart
Notes:
• Relative frequencies may be used instead.
• All rectangles must have the same width and not overlap.
• If the categories are integer values, centre the rectangles

on the category values.
Bar chart: Example.
Ridge hospital ward experienced the following distribution

on the number of deaths each month over a 47 month period.
• Number of deaths per month 0 1 2 3 4 5
• Number of such months 6 11 16 9 4 1

Bar chart: Example.
Solution
Pie chart
Suitable for qualitative (categorical) data: data values

associated with a small number of categories. Areas reflect
these values.
• Example: consider the following data set (coloured balls):
Red Orange Yellow Green Blue
9 4 12 21 17
Pie chart
• Assuming a circle of radius r = 1, what would be the area

associated with the Green slice in a pie chart? The total
count is 63,
21 1
• so the proportion of pie area for Green is = . Hence,
63 3
𝜋𝑟 2
the area for Green is = 1.047198.
3
Pie and bar charts
Bar chart is much better than pie charts for displaying

categorical data:
• relative lengths (of bars) are easier to judge than relative

areas (in a pie chart)
• separate bars represent distinct values observed:
• bar heights are proportional to the frequency (number of
observations)
Stem and leaf plots.
• Useful for a small set of numeric data.
• Gives an impression of location, spread and shape of

values.
• More usually there is only 1 digit in a leaf value, in which

case do not use spaces to separate values.
Steps to construct Stem-and-Leaf
Plots.
1. Select one or more leading digits to make the Stem
values. The remaining digits become the leaves.
2. List possible stem values vertically.
3. Record each observation by its leaf value in the

corresponding stem row.
Steps to construct Stem-and-Leaf
Plots.
4. Indicate the stem and leaf units nearby.
5. If there are some large or small isolated values, list those

separately near the stem plot.
6. Optionally, record the frequency or cumulative frequency

in another column.
For ordered stem-and-leaf plots, sort the leaf values in each

row.
Example.
Consider 20 claim sizes on a certain insurance policy.
• 54 127 700 250 81 17 90 28 1100 320

• 547 14 38 51 130 31 148 600 130 200
solution
Stem and leaf plot (for previous data)
Soln.
Variations of Stem-and-Leaf plots
• There is no unique convention for drawing stem-and-leaf

plots, you will find a range of conventions in different
textbooks.
• Some insist on having a cumulative frequency column; in

some the stems will increase going up, in others going
down.
Variations of Stem-and-Leaf plots
• There are variations that open up' the stems if the majority
of the observations are in one or two leaves.
• You should be aware of these variations in case you come

across them, but computer packages can output them
automatically so we will not go into their hand
construction here.

STAT 111: Introduction To Statistics and Probability: Lecture 2: Data Reduction

Uploaded by

Copyright:

Available Formats

STAT 111: Introduction To Statistics and Probability: Lecture 2: Data Reduction

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

STAT 111: Introduction To Statistics and Probability: Lecture 2: Data Reduction

Uploaded by

Copyright:

Available Formats

STAT 111: Introduction to Statistics

Lecture 2: Data Reduction

Department of Statistics and Actuarial Science

 Populations and samples

 Data types: qualitative versus quantitative

 Data displays: pie chart, bar chart, stem-and-leaf

 Data numerical summaries: sample mean, sample

Sample is a subset taken from a larger population.

• Example: all the students in this lecture theatre form a

Elements in the sample are called sampling units.

• Example above: students are the sampling units.

• Random sample is a collection of sampling units chosen

• If selected unfairly, then it is called biased sample.

Example: in an opinion poll a predetermined proportion is

 Qualitative (descriptive, non-numerical)

• Nominal - named, unordered categories

• Ordinal - ordered categories

• Discrete - distinct values (usually integer values)

• Continuous - any real value in a range

Data consists of a collection of observations (e.g., heights of

students in this lecture theatre).

• A variate is qualitative if there is no meaningful way to

• collect autumn leaves; the variate can be the species of

• observe vehicles passing a census point and record their

• Example: look at the next 20 buses to go past and record

• feedback forms have a 5-point scale from ”Very poor" to

• Displays should be appropriate, have impact and be

• Both useful for categorical data with a few number of

• In a pie chart, the areas of a segment represents the

• In a bar chart, the height of a bar represents the value.

• Generally bar charts are better for display. Relative lengths

1. Draw a horizontal line with category names at regular

2. Draw a vertical axis scaled to the largest frequency

3. Above each category draw a rectangle with height at the

• Relative frequencies may be used instead.

• If the categories are integer values, centre the rectangles

Ridge hospital ward experienced the following distribution

• Number of deaths per month 0 1 2 3 4 5

• Number of such months 6 11 16 9 4 1

Suitable for qualitative (categorical) data: data values

• Example: consider the following data set (coloured balls):

Red Orange Yellow Green Blue

• Assuming a circle of radius r = 1, what would be the area

Bar chart is much better than pie charts for displaying

• relative lengths (of bars) are easier to judge than relative

• separate bars represent distinct values observed:

• bar heights are proportional to the frequency (number of

• Useful for a small set of numeric data.

• Gives an impression of location, spread and shape of

• More usually there is only 1 digit in a leaf value, in which

2. List possible stem values vertically.

3. Record each observation by its leaf value in the

5. If there are some large or small isolated values, list those

6. Optionally, record the frequency or cumulative frequency

For ordered stem-and-leaf plots, sort the leaf values in each

Consider 20 claim sizes on a certain insurance policy.

• 54 127 700 250 81 17 90 28 1100 320

• There is no unique convention for drawing stem-and-leaf

• Some insist on having a cumulative frequency column; in