Why Are Data Collected?: STAT 235 - 1 2 / 31

Why are data collected?
STAT 235 | 1
2 / 31
K
What kind of data are collected?
STAT 235 | 1
3 / 31
K
❑ Medical & pharmaceutical sciences...
STAT 235 | 1
3 / 31
K
❑Medical & pharmaceutical sciences...

❑ Financial companies & institutions...
STAT 235 | 1
3 / 31
K

❑ Engineering & computer science...
STAT 235 | 1
3 / 31
K

❑ Social & behavioral sciences...
STAT 235 | 1
3 / 31
K

❑ Education & research...
STAT 235 | 1
3 / 31
K

❑ Sport, entertainment & fun...
STAT 235 | 1
3 / 31
K

❑ Sport, entertainment & fun...
❑ etc.
STAT 235 | 1
3 / 31
K
STAT 235 - Introduction
How to do it in a proper way?
STAT 235 | 1
4 / 31
K
Syllabus
❑ Introduction & Descriptive Statistics chapter 6

❑ Introduction to Probability chapter 1
❑ Random Variables chapter 2
❑ Discrete and Continuous Probability Distributions chapters 3,4
❑ The Normal Probability Distribution chapter 5
❑ Sampling Distributions, Random Sample chapter 7
❑ Inferences on a Population Mean chapter 8
❑ Comparing Two Population Means chapter 9
❑ Simple Linear Regression and Correlation chapter 12
❑ Inferences on a Population Proportion chapter 10
❑ The Analysis of Variance chapter 11
STAT 235 | 1
5 / 31
K
STAT 235 | 1
6 / 31
K
Probability ←v
Statistics →s.
❑ Random mechanism → produces random
outcomes;
STAT 235 | 1
6 / 31
K
Probability ←v
Statistics →s.
❑ Random mechanism → produces random outcomes;
❑ A set of random outcomes (results) → drawing
conclusions;
STAT 235 | 1
6 / 31
K
Probability ←v
Statistics →s.
❑ A set of random outcomes (results) → drawing conclusions;
❑ Probability - theoretical background for a random mechanism;

(describes how the mechanism works - but the principle is unknown)
STAT 235 | 1
6 / 31
K
Probability ←v
Statistics →s.
❑ A set of random outcomes (results) → drawing conclusions;
❑ Probability - theoretical background for a random mechanism;

(describes how the mechanism works - but the principle is unknown)
❑ Statistics - uses random outcomes to draw conclusions;
(using statistics we can very precisely describe the mechanism behind)
STAT 235 | 1
6 / 31
K
Statistics in Real Life...
❑ In probability, we could have aprecise description of the random

mechanism behind whichwouldbe accurate, but it is unknown!
❑ In statistics, we have data (measurements) that are quite often not

precise, but the final conclusions drawn from such data are very accurate!
STAT 235 | 1
7 / 31
K
Statistics in Real Life...
❑ In probability, we could have aprecise description of the random

mechanism behind whichwouldbe accurate, but it is unknown!
❑ In statistics, we have data (measurements) that are quite often not

precise, but the final conclusions drawn from such data are very accurate!
❑ “I only believe in statistics that I doctored myself.”

Winston S. Churchill (1874 – 1965)
STAT 235 | 1
7 / 31
K
Population vs. Sample
STAT 235 | 1
8 / 31
K
Probability Theory
Random Distribution
STAT 235 | 1
8 / 31
K
Probability Theory
Random Distribution
STAT 235 | 1
8 / 31
K
Probability Theory Finite Sample
Random Distribution Empirical Distribution
STAT 235 | 1
8 / 31
K
Probability Theory Finite Sample
STAT 235 | 1
8 / 31
K
Probability Theory Finite Sample Sample Statistics
STAT 235 | 1
8 / 31
K
Probability Theory Finite Sample Sample Statistics

Random Distribution Empirical Distribution Statistical Inference
STAT 235 | 1
8 / 31
K
Different Data types
❑ Categorical variables
STAT 235 | 1
9 / 31
K
❑ Numerical variables
STAT 235 | 1
9 / 31
K
❑ Nominal categories
❑ categories with no ordering;
What kind of transportation do you use to get to work?
What kind of material do you mostly prefer?
STAT 235 | 1
9 / 31
K
❑ Ordinal categories
❑ categories with possible ordering;
What was your most frequent grade last semester?
What energy category does the machine belong to?
STAT 235 | 1
9 / 31
K
❑ Integer values - counts
❑ an apriori assumption of equidistant differences;
How many people visit this museum a day? How
many call a day is addressed to 911?
STAT 235 | 1
9 / 31
K
❑ Integer values - counts
❑ an apriori assumption of equidistant differences;
How many people visit this museum a day? How
many call a day is addressed to 911?
❑ Real values
❑ any real value is possible...
length, weight, height, temperature, etc.
STAT 235 | 1
9 / 31
K
Data exploration
❑ What is the nature of data?

❑ What are the limitations of the experiment
behind?
❑ What is the main question of interest?

❑ Is it possible to use data available to answer it?
❑ Is it important to visualize the data?

❑ Howcan one get an insight in the data?
STAT 235 | 1
10 / 31
K
Binary Data
❑ the simplest data type...
❑ different coding possible...
❑ logical:TRUE | FALSE; 1|0; YES|NO;
❑ categorical:A|B; 1|2; HOME|ABROAD;
❑ only afew options on how to represent such data;
STAT 235 | 1
11 / 31
K
Binary Data
Summary
(table):
TRUE
41
FALSE
59
STAT 235 | 1
11 / 31
K
Binary Data
Summary Frequency
(table): summary:
TRUE
TRUE: 41.00 %
41
FALSE
59
STAT 235 | 1
11 / 31
K
Binary Data
Summary Frequency Graphical

(table): summary: view:
TRUE
TRUE: 41.00 %
41
0 10 20 3 0 40 50
FALSE FALSE TRUE
59
STAT 235 | 1
11 / 31
K
Categorical Data
❑ Nominal categories...
❑ Machine Breakdowns: what is a machine breakdown cause?
STAT 235 | 1
12 / 31
K
Categorical Data
elektrical | mechanical | misuse
STAT 235 | 1
12 / 31
K
Categorical Data
❑ Data: a sequence of words’electrical’, ’mechanical’or’misuse’;
STAT 235 | 1
12 / 31
K
Categorical Data
Summary
(table):
Electrical 9
Mechanical 24
Misuse 13
STAT 235 | 1
12 / 31
K
Categorical Data
Summary Frequency
(table): summary:
Electrical 9 Electrical 19.57 %

Mechanical 24 Mechanical 52.17 %
Misuse 13 Misuse 28.26 %
STAT 235 | 1
12 / 31
K
Categorical Data


0 5 10 1 5 2 0
electrical mechanical misuse
STAT 235 | 1
12 / 31
K
Categorical Data


STAT 235 | 1
12 / 31
K
Categorical Data
❑ Ordinal categories...
❑ Number of Breakdowns: How many times did a machine break down?
STAT 235 | 1
13 / 31
K
Categorical Data
0 times | once | two times and more
STAT 235 | 1
13 / 31
K
Categorical Data
❑ Data: a sequence of values0,1and2;
STAT 235 | 1
13 / 31
K
Categorical Data
Summary
(table):
Zero 54
Once 40
≥2 6
STAT 235 | 1
13 / 31
K
Categorical Data
Summary Frequency
(table): summary:
Zero 54 Zero 54 %
Once 40 Once 40 %
≥2 6 ≥2 6%
STAT 235 | 1
13 / 31
K
Categorical Data

Zero 54 Zero 54 %
Once 40 Once 40 %
50
≥2 6 ≥2 6%
0 1 0 2 0 3 0 40
>= 2 once zero
STAT 235 | 1
13 / 31
K
Categorical Data

Zero 54 Zero 54 %
Once 40 Once 40 %
≥2 6 ≥2 6%
STAT 235 | 1
13 / 31
K
Categorical Data

Zero 54 Zero 54 %
Once 40 Once 40 %
≥2 6 ≥2 6%
❑ What is the main difference between nominal and ordinal categories?
STAT 235 | 1
13 / 31
K
Numerical Data
❑ Integer values - counts...
❑ How many hours cana machine work until it breaks down?
STAT 235 | 1
14 / 31
K
Numerical Data
0, 1, 2, 3, ... etc.
STAT 235 | 1
14 / 31
K
Numerical Data
0, 1, 2, 3, ... etc.
❑ How many people a day will use the lift?
0, 1, 2, 3, ... etc.
STAT 235 | 1
14 / 31
K
Numerical Data
0, 1, 2, 3, ... etc.
0, 1, 2, 3, ... etc.
❑ How many passengers is in a driving car?
STAT 235 | 1
14 / 31
K
Numerical Data
0, 1, 2, 3, ... etc.
0, 1, 2, 3, ... etc.
1, 2, 3, etc. ≤ 5
STAT 235 | 1
14 / 31
K
Numerical Data
0, 1, 2, 3, ... etc.
0, 1, 2, 3, ... etc.
1, 2, 3, etc. ≤ 10
STAT 235 | 1
14 / 31
K
Numerical Data
0, 1, 2, 3, ... etc.
0, 1, 2, 3, ... etc.
1, 2, 3, etc. ≤ 30
STAT 235 | 1
14 / 31
K
Numerical Data
0, 1, 2, 3, ... etc.
0, 1, 2, 3, ... etc.
1, 2, 3, etc. ≤ 100
STAT 235 | 1
14 / 31
K
Numerical Data
0, 1, 2, 3, ... etc.
0, 1, 2, 3, ... etc.
1, 2, 3, etc. ≤ 100
❑ Data: mostly a sequence of integer values...
STAT 235 | 1
14 / 31
K
Numerical Data
0, 1, 2, 3, ... etc.
0, 1, 2, 3, ... etc.
1, 2, 3, etc. ≤ 100
15
10
table(rr)
5
0
4 5 6 7 8 9 10 11 12 13 14 15 16 18
STAT 235 | 1
14 / 31
K
Numerical Data
0, 1, 2, 3, ... etc.
0, 1, 2, 3, ... etc.
1, 2, 3, etc. ≤ 100
15
0.08
table(rr)
De ns i ty
0 . 00 0. 0 2 0 .0 4 0 . 06
10
5
0
1 6 5 1 70 17 5 1 8 0 1 8 5 19 0 1 95
4 5 6 7 8 9 10 11 12 13 14 15 16 18 Hei g ht o f m a l e s
STAT 235 | 1
14 / 31
K
Numerical Data
0, 1, 2, 3, ... etc.
0, 1, 2, 3, ... etc.
1, 2, 3, etc. ≤ 100
❑ Data: mostly a sequence of integer values... 100
10
15
table(rr)
F re q u e n c y
150
50
5
0
0
1 6 5 1 70 17 5 1 8 0 1 8 5 19 0 1 95
4 5 6 7 8 9 10 11 12 13 14 15 16 18 Hei g ht o f m a l e s
STAT 235 | 1
14 / 31
K
Numerical Data
0, 1, 2, 3, ... etc.
0, 1, 2, 3, ... etc.
1, 2, 3, etc. ≤ 100
15
150
10
F re q u e n c y
table(rr)
4 6 8 10 12 14 16 18
50
5
100
0
0
1 6 5 1 70 17 5 1 8 0 1 8 5 19 0 1 95
4 5 6 7 8 9 10 11 12 13 14 15 16 18 Hei g ht o f m a l e s
STAT 235 | 1
14 / 31
K
Numerical Data
❑ Real values...
❑ We could improve a precision on the previous example...
STAT 235 | 1
15 / 31
K
Numerical Data
❑ Real values...
data → positive real values...
STAT 235 | 1
15 / 31
K
Numerical Data
❑ Real values...
❑ How long cana machine work until it breaks down?
STAT 235 | 1
15 / 31
K
Numerical Data
❑ Real values...
❑ Data: a sequence of real valued observations...
❑ infinitely many different outcomes to be considered;
❑ the most common data presentation using
histograms;
STAT 235 | 1
15 / 31
K
Numerical Data
❑ Real values...
histograms;
0.10
0.10
0.08
0.08
0.06
Density
Density
0.04 0.06
0.04
0.02
0.02
0.00
0.00
165 170 175 180 185 190 195 155 160 165 170 175 180 185 190
Height of males Height of females
STAT 235 | 1
15 / 31
K
Numerical Data
❑ Real values...
histograms;
0.10


195


0.08
190
185
Density
0.04 0.06
180
175
170
0.00 0.02


155 160 165 170 175 180 185 190
Height of females
STAT 235 | 1
15 / 31
K
Numerical Data
❑ Real values...
histograms;
 
 
195
 

185
190
180
185
175
180
170
175
165
170
160

 
STAT 235 | 1
15 / 31
K
Numerical Data
❑ Real values...
histograms;
0.10



0.08
175 180 185

0.06
Density
170
0.04
160 165

0.00 0.02
165 170 175 180 185 190 195
Height of males
STAT 235 | 1
15 / 31
K
Numerical Data
❑ Real values...
histograms;
0.10
0.10
0.08
0.08
0.06
Density
Density
0.04 0.06
0.04
0.02
0.02
0.00
0.00
165 170 175 180 185 190 195 155 160 165 170 175 180 185 190
STAT 235 | 1
15 / 31
K
Numerical Data
❑ Real values...
histograms;
0.10
0.10
0.08
0.08
0.06
Density
Density
0.04 0.06
0.04
0.02
0.02
0.00
0.00
165 170 175 180 185 190 195 155 160 165 170 175 180 185 190
STAT 235 | 1
15 / 31
K
Numerical Data
❑ Real values...
histograms;
0.10
0.10
0.08
0.08
0.06
Density
Density
0.04 0.06
0.04
0.02
0.02
0.00
0.00
165 170 175 180 185 190 195 155 160 165 170 175 180 185 190
STAT 235 | 1
15 / 31
K
Histogram...
0.08
0.06
Density
0.04
0.02
0.00
165 170 175 180 185 190 195
Height of males
STAT 235 | 1
16 / 31
K
Histogram...
150
100
Frequency
50
0
165 170 175 180 185 190 195
Height of males
STAT 235 | 1
16 / 31
K
Histogram...
300
Frequency
200
100
0
165 170 175 180 185 190 195 200
Height of males
STAT 235 | 1
16 / 31
K
Histogram...
40
30
Frequency
20
10
0
170 175 180 185 190 195
Height of males
STAT 235 | 1
16 / 31
K
Histogram...
15
10
Frequency
5
0
170 175 180 185 190 195
Height of males
STAT 235 | 1
16 / 31
K
Boxplot figure...
STAT 235 | 1
17 / 31
K
Some nice properties
What can be directly observed from histogram (boxplot)?
❑ symmetric/non-symmetric data distribution;

❑ unimodal or multimodal data distribution;
❑ skewness of data distribution;
❑ indications of outlying observations;
❑ ...
STAT 235 | 1
18 / 31
K
Some nice properties
What can be directly observed from histogram (boxplot)?
❑ symmetric/non-symmetric data distribution;

❑ unimodal or multimodal data distribution;
❑ skewness of data distribution;
❑ indications of outlying observations;
❑ ...
It is not so much important now when we only focus on data exploration

part however, it will become really crucial when considering more
sophisticated statistical modeling approaches.
STAT 235 | 1
18 / 31
K
Data sample reliability
❑ How reliable such data samples are?

❑ Could we expect more reliability and confidence if more data is available?
STAT 235 | 1
19 / 31
K

❑ Is there a way to numerically express such reliability, confidence?

❑ What should be a sufficient number of observation?
STAT 235 | 1
19 / 31
K

❑ Is there a way to numerically express such reliability, confidence?

❑ What should be a sufficient number of observation?
❑ The more data we obtain the more we have to pay for...

❑ How to find a reasonable balance?
STAT 235 | 1
19 / 31
K
From counts to ordinal categories

❑ How long does the machine work until it breaks down?
STAT 235 | 1
20 / 31
K

❑ more that 15 hours ⇒ category A
STAT 235 | 1
20 / 31
K

❑ more that 10 & less than 15 hours ⇒ category B
STAT 235 | 1
20 / 31
K

❑ more that 5 & less than 10 hours ⇒ category C
STAT 235 | 1
20 / 31
K

❑ less than 5 hours ⇒ category D
STAT 235 | 1
20 / 31
K

❑ What is the data sample now?
STAT 235 | 1
20 / 31
K

❑ What is the data sample now?

10 20 30 40 50
0
category A category B category C category D
STAT 235 | 1
20 / 31
K
Higher dimensional data
❑ two categorical variables;

❑ one categorical and the other numerical variable;
❑ two numerical (continuous) variables;
STAT 235 | 1
21 / 31
K
Higher dimensional data
❑ two categorical variables;

❑ one categorical and the other numerical variable;
❑ two numerical (continuous) variables;
❑ more dimensional data ⇒ it gets even difficult to plot;

❑ multivariate statistical methods were proposed to be used
instead;
STAT 235 | 1
21 / 31
K
Titanic survivals
Sex Sex
C lass Male C lass Male
1st
Female 0 0 1st
Female 5 1
2nd 0 0 2nd 11 13
3rd 35 17 3rd 13 14
Crew 0 0 Crew 0 0
, , Age = C h i l d , Survived = No , , Age = C h i l d , Survived = Yes
Sex Sex
C lass Male C lass Male Female
1st
Female 118 4 1st 57
2nd 154 13 2nd 14140 80
3rd 387 89 3rd 75 76
Crew 670 3 Crew 192 20
, , Age = A d u l t , Survived = No , , Age = A d u l t , Survived = Yes
STAT 235 | 1
22 / 31
K
Car Breaking Distance

120

100
 
80






dist
60

 
 



 

40
 
 
 
  
 
   


20
 
 


 


0
5 10 15 20 25
speed
STAT 235 | 1
23 / 31
K
Orange Trees
STAT 235 | 1
24 / 31
K
Sampling procedure
From Population to its Sample
population =⇒ population sample =⇒ statistical inference
STAT 235 | 1
25 / 31
K
Sampling procedure
Sampling procedure:
❑ ideally we would like to obtain arandom
sample;
STAT 235 | 1
25 / 31
K
Sampling procedure
Sampling procedure:
❑ ideally we would like to obtain arandom sample;
❑ random sample → independent and identically distributed
observations;
STAT 235 | 1
25 / 31
K
Sample Statistics
STAT 235 | 1
26 / 31
K
Sample Statistics
Two different worlds
...
STAT 235 | 1
27 / 31
K
Sample Statistics
Sample Mean
❑ sample mean ≡ average
STAT 235 | 1
28 / 31
K
Sample Statistics
Sample Mean
❑ sample mean ≡ average ƒ= mean
STAT 235 | 1
28 / 31
K
Sample Statistics
Sample Mean
❑ arithmetic average of all given observations;

Σ n
{x1, x2, . . . , xn
i= 1
❑notation: x n = n
xi
, wherethe actual observations are
};
STAT 235 | 1
28 / 31
K
Sample Statistics
Sample Mean

Σ n
{x1, x2, . . . , xn
i= 1
xi
❑ it is a “middle value” with respect tovalues and their distribution;
};
STAT 235 | 1
28 / 31
K
Sample Statistics
Sample Mean

Σ n
{x1, x2, . . . , xn
i= 1
xi
};
STAT 235 | 1
28 / 31
K
Sample Statistics
Sample Mean

Σ n
{x1, x2, . . . , xn
i= 1
xi
};
❑ Advantage:it is sensitive with respect to outlying observations;
STAT 235 | 1
28 / 31
K
Sample Statistics
Sample Mean

Σ n
{x1, x2, . . . , xn
i= 1
xi
};

❑ Disadvantage:it issensitive with respect to outlying observations;
STAT 235 | 1
28 / 31
K
Sample Statistics
Sample Mean

Σ n
{x1, x2, . . . , xn
i= 1
xi
};

❑ Disadvantage:it issensitive with respect to outlying observations;
❑ Some other proposals: sample trimmed mean, weighted mean, etc.;
STAT 235 | 1
28 / 31
K
Sample Statistics
Sample Median
❑ middle observation from all ordered observations;

❑ x˜n= x( n+1 ) for n even and x˜n= [x(n/2) + x( n+1 )]/2 for n odd;
2 2
STAT 235 | 1
29 / 31
K
Sample Statistics
Sample Median

2 2
❑ it is a “middle value” with respect to

givenvaluesthemselves;
STAT 235 | 1
29 / 31
K
Sample Statistics
Sample Median

2 2

STAT 235 | 1
29 / 31
K
Sample Statistics
Sample Median

2 2

❑ Advantage:it is not sensitive with respect to observation

values;
STAT 235 | 1
29 / 31
K
Sample Statistics
Sample Median

2 2
❑ it is a “middle value” with respect to givenvaluesthemselves;
❑ Advantage:it is not sensitive with respect to observation values;

❑ Disadvantage:it is not sensitive with respect to observations
values;
STAT 235 | 1
29 / 31
K
Sample Statistics
Sample Median

2 2
❑ it is a “middle value” with respect to givenvaluesthemselves;
❑ Advantage:it is not sensitive with respect to observation values;

❑ Disadvantage:it is not sensitive with respect to observations
values;
❑ Some other proposals as well ...;
STAT 235 | 1
29 / 31
K
Sample Statistics
Some other sample characteristics
❑ sample mode (for categorical variables mostly);

What is the relation between sample mean, median and mode?
❑ sample variance & sample standard error;
mostly it is difficult to imagine → sample range used “instead”
❑ sample quantiles (quartiles, percentiles, etc.);
some of them are more important than others...
❑ coefficient of variation;
spread of data relative to its middle value
STAT 235 | 1
30 / 31
K
Sample Statistics
To be continued...
❑ Probability, probability concepts;

❑ Random events, combination of events;
❑ Conditional probability;
❑ Independence principle;
❑ Law of Total Probability;
❑ Bayes Theorem in Probability;
STAT 235 | 1
31 / 31
K

Why Are Data Collected?: STAT 235 - 1 2 / 31

Diunggah oleh

Informasi Dokumen

Deskripsi Asli:

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Why Are Data Collected?: STAT 235 - 1 2 / 31

Diunggah oleh

Hak Cipta:

Format Tersedia

Why are data collected?

❑ Medical & pharmaceutical sciences...

❑Medical & pharmaceutical sciences...

❑Medical & pharmaceutical sciences...

❑Medical & pharmaceutical sciences...

❑Medical & pharmaceutical sciences...

❑Medical & pharmaceutical sciences...

❑Medical & pharmaceutical sciences...

How to do it in a proper way?

❑ Introduction & Descriptive Statistics chapter 6

How to do it in a proper way?

How to do it in a proper way?

How to do it in a proper way?

How to do it in a proper way?

❑ Probability - theoretical background for a random mechanism;

How to do it in a proper way?

❑ Probability - theoretical background for a random mechanism;

Statistics in Real Life...

❑ In probability, we could have aprecise description of the random

❑ In statistics, we have data (measurements) that are quite often not

Statistics in Real Life...

❑ In probability, we could have aprecise description of the random

❑ In statistics, we have data (measurements) that are quite often not

❑ “I only believe in statistics that I doctored myself.”

Population vs. Sample

Population vs. Sample

Population vs. Sample

Population vs. Sample

Probability Theory Finite Sample

Random Distribution Empirical Distribution

Population vs. Sample

Probability Theory Finite Sample

Random Distribution Empirical Distribution

Population vs. Sample

Probability Theory Finite Sample Sample Statistics

Random Distribution Empirical Distribution

Population vs. Sample

Probability Theory Finite Sample Sample Statistics

Different Data types

Different Data types

Different Data types

Different Data types

Different Data types

Different Data types

❑ What is the nature of data?

❑ What is the main question of interest?

❑ Is it important to visualize the data?

Summary Frequency Graphical

Electrical 9 Electrical 19.57 %

Summary Frequency Graphical

Electrical 9 Electrical 19.57 %

Summary Frequency Graphical

Electrical 9 Electrical 19.57 %

Summary Frequency Graphical

Summary Frequency Graphical

Summary Frequency Graphical

❑ What is the main difference between nominal and ordinal categories?

Height of males Height of females

155 160 165 170 175 180 185 190

175 180 185

165 170 175 180 185 190 195

Height of males Height of females