ENGINEERING DATA ANALYSIS
Roles of Statistics and the Data Analysis Process
Statistics
 defined as the branch of science that deals with
collection, presentation, organization, and
interpretation of data
 widely used in many different fields of science
and technology, as well as in our everyday lives.
Examples of applicants of statistics include (but not
limited to);
 Population census (as implemented by the
government);
 Public choices/ responses (e.g., surveys related
to elections and government service
satisfaction);
 Product advertisements (e.g., comparison of
Brand X vs Brand Y);
 Teaching and instruction (e.g., student’s
performance, examination item analysis etc.)
 Scientific observations and experiments (e.g.,
clinical trials for medicines and vaccines,
development of new technologies etc.); and
 Engineering data collection (e.g., engineering
soil properties, testing of materials, etc.)
Uncertainty and Variability
Uncertainty
 Occurs when the true value of a certain
quantity at a single instance is unknown
 Uncertainty is derived from theoretical
information (i.e., it is expressed in terms of
probabilities)
Variability
 Occurs when the said quantity is measured at
multiple instances and there are considerable
differences between the said measurement
trials.
 Derived form data extracted from observations
and experiments (i.e., it is expressed in terms of
frequencies).
 Uncertainty emerges because of variability
SOURCES OF UNCERTAINTY
Aleatory uncertainties
 those are caused by natural randomness. It is
natural in reality that the properties of several
samples of a certain object may be different.
 An example of occurrence of aleatory
uncertainty is determination of compressive
strength of sandstone from a single source. If
experimentation is done, assuming that there
are no mistakes in experiment set-up, it can be
observed that for every sample, different values
of strength may come up.
Epistemic Uncertainties
 Those that are caused by an “incomplete”
understanding of reality. For simplicity of
understanding of some physical
phenomena, assumptions must be made.
However, these assumptions may not
capture what is actually happening in
reality.
 An example of epistemic uncertainty is
determination of beam deflections. An
important assumption in calculating beam
deflections is that the beam material must
be perfectly elastic. However, most
materials are elastic at certain stresses only;
meaning, at some large magnitude of load
applied on the beam, the beam material
may behave differently from that of a
perfectly elastic material. If
experimentation is done for this, it is
expected that there will be a difference
between the expected value (calculated
value) and the observed value (measured
value).
DATA ANALYSIS PROCESS
1. Understanding the nature of the problem.
Effective data analysis requires an
understanding of the research problem. We
must know the goal of the research and what
questions we hope to answer. It is important to
have a clear direction before gathering data to
ensure that we will be able to answer the
questions of interest using the data collected.
2. Deciding what to measure and how to
measure it. The next step in the process is
deciding what information is needed to answer
the questions of interest. In some cases, the
choice is obvious.
3. Data collection. The data collection step is very
important. The researcher must first decide
whether an existing data source is adequate or
whether new data must be collected. If a
decision is made to use existing data, it is
important to understand how the data were
collected and for what purpose, so that any
resulting limitations are also fully understood. If
new data are to be collected, a careful plan
 must be developed, because the type of
analysis that is appropriate and the conclusions
that can be drawn depend on how the data are
collected.
4. Data summarization and preliminary analysis.
After the data are collected, the next step is
usually a preliminary analysis that includes
summarizing the data graphically and
numerically. This initial analysis provides insight
into important characteristics of the data and
provides guidance in selecting appropriate
methods for further analysis.
5. Formal data analysis. The data analysis step
requires the researcher to select appropriate
statistical methods.
6. Interpretation of results. The interpretation
step often leads to the formulation of new
research questions. These new questions lead
back to the first step. In this way, good data
analysis is often an iterative process.
 A population refers to the entire collection of
individuals or objects about which information
is desired, while a sample is a representation of
the population where this group comes from,
i.e., making it a subset of a population.
 A summary measure that describes a specific
characteristic of a sample is called a statistic,
while a summary measure that describes a
specific characteristic of a population is called a
parameter. This means that the data obtained
from descriptive statistics are examples of a
statistic while the data obtained from
inferential statistics are examples of a
parameter.
 Designed research that provides information
needed to solve a certain research problem is
called a statistical inquiry.
DATA
 A data is a collection of observations on one or
more variables.
 A variable is a characteristic whose value may
change from one observation to another.
CLASSIFICATION OF DATA
 Data may be depending on its nature and on
the number of involved variables. Depending on
its nature, data may be classified as either:
1. Categorical (qualitative): if the individual
observations are categorical responses; or
2. Numerical (quantitative): if the individual
observations are expressed as numbers.
 Numerical data may be further classified as
follows:
1. Discrete: if possible values of the variable/s
correspond to isolated points on the
number line; and
2. Continuous: if possible values of the
variable/s correspond to all points inside an
interval on the number line.
 Depending on the number of variables involved,
data may be classified as either:
1. Univariate: if a data set consists
observations on a single variable; or
2. Multivariate: if a data set consists
observations on two or more variables.
DATA COLLECTION & SAMPLING TECHNIQUES
Data Collection Methods
 Use of Documented Data: Available research
data may be used such as government data and
data from related studies or researches.
 However, caution must be exercised when
using documented data, especially
secondary data (i.e., data documented by
entities other than the actual data
collection).
 Surveys: A survey is a method of collecting data
on the variable of interest by asking people to
answer a set of carefully written questions
called a questionnaire.
 A survey comprising an entire population is
called a census while a survey comprising
only a sample of the population is called a
sample survey.
 Surveys are usually performed if the study
involves human behavior such as consumer
studies and election surveys.
 Experiments: An experiment is a method of
collecting data where there is a direct human
intervention on the conditions that may affect
the values of the variable of interest.
 Variables that may be directly manipulated
are called independent variables while
variables that cannot be manipulated but
can have their values changed are called
dependent variables.
 Most scientific studies with multivariate
data involve experimentation.
  Observations: An observation is a method of
collecting data on the phenomenon of interest
by recording the observations made about the
phenomenon as it actually happens.
 Examples of studies involving observations
include weather and climate, earthquake,
and astronomical studies.
Sampling
The process of obtaining or selecting samples from a
population related to a study is called sampling.
DESIGN OF EXPERIMENTS
Sampling Bias
 Sampling must be carefully performed for
improper sampling may cause bias, which is the
tendency for samples to differ from the
corresponding population in some systematic
way.
 Bias results either from the sampling itself or
from the way in which data is obtained once the
sample was chosen.
 The three most common types of bias in
sampling are as follows:
 Selection Bias: Tendency for samples to
differ from the population as a result of a
systematic exclusion of some part of the
population.
 Measurement or Response Bias: Tendency
for samples to differ from the population
because the method of observation tends
to produce values that differ from the true
values.
 Nonresponse Bias: Tendency for samples to
differ from the population because data are
not obtained from all individuals selected
for inclusion in the sample.
Sampling Methods
 Random Sampling: It is a sampling method that
ensures that every different sample of a certain
size has an equal chance of being chosen as the
sample. Random sampling may be done as
either sampling without replacement (once
chosen, cannot be chosen again) or sampling
with replacement (once chosen, may not be
chosen again).
 Stratified Random Sampling: It is a type of
sampling method that divides the population
into a set of non-overlapping subgroups, and
then random sampling is done for each of these
subgroups.
 Cluster Sampling: It is a type of sampling
method that involves dividing the population of
interest into non-overlapping subgroups called
clusters, and then these clusters are selected at
random, with all individuals in the selected
clusters are included in the sample.
 Systematic Sampling: It is a sampling method
that can be used when it is possible to view the
population of interest as consisting of a list or
some other sequential arrangement.
 An experiment is a method of collecting data
where there is a direct human intervention on
the conditions that may affect the values of the
variable of interest.
 It is a study in which one or more
explanatory variables are manipulated in
order to observe the effect on a response
variable.
 Explanatory variables are independent
variables or factors, those that have values that
are controlled by the experimenter
 Response variables are dependent variables,
those that are thought to be related to the
explanatory variables in an experiment. These
are measured as part of the experiment, but
not controlled by the experimenter.
 An experimental study involves several set-ups,
called experimental conditions or treatments,
to observe the relationship between the
independent and the dependent variables.
 The main goal of an experiment is to determine
the effects of independent variables on the
dependent variables.
 A well-designed experiment requires not just
manipulating the independent variables, but
also eliminating the effects of other variables
not involved in the study on the dependent
variables.
 These variables that are not included as
independent variables but may affect the
dependent variables are called extraneous
variables.
 If extraneous variables are left alone, then
the independent variables may be
confounded, i.e., if their effects on the
dependent variable cannot be distinguished
from one another.
Strategies for design of experiments which can control
the effect of extraneous variables may be employed, as
follows:
 Random Assignment: Random assignment (of
subjects to treatments or of treatments to
trials) to ensure that the experiment does not
systematically favor one experimental condition
(treatment) over another.
 Blocking: Using extraneous variables to create
groups (blocks) that are similar. All
experimental conditions (treatments) are then
tried in each block.
 Direct Control: Holding extraneous variables so
that their effects are not confounded with those
of the experimental conditions (treatments).
 Replication: Ensuring that there is an adequate
number of observations for each experimental
condition.
DATA PRESENTATION & ORGANIZATION
Excerpt taken from the business section of the
Philippine Star:
1. “The 30-company Philippine Stock Exchange
Index finished down 10.21 points, or 0.5
percent, at 1921.33, after failing 0.8 percent
Tuesday following a seven-day rally that
boosted the main index by 6.4 percent.
Weighing on the index were losses incurred by GRAPHICAL
Globe Telecom, down by 2.4 percent at Php 830 Portrays numerical figures or relationships among
Ayala Land, off 1.2 percent at php 8, and variables in pictorial form.
Jollibee, lower by 3.4 percent at Php 28.50 on Raw Dara and Array: Raw data are data in their original
profit taking.” form while array is an ordered arrangement of data
according to magnitude (also called sorted data or
2. “Partly offsetting the market’s losses were Ayala ordered data).
Corporation, up 1.4 percent at 6.30, and Bank of
the Philippine Islands, which rose one percent
at Php 50 on bargain hunting. Ayala unit Manila
Water rose 1.7 percent to Php 6.10 after the
water utility Tuesday posted a 57-percent year-
on-year rise in first -quarter net profit.”

3. “All sectoral indicators ended lower, except the

oil sub index, which finished higher. Declines led
gainers 46 to 21, while 51 stocks were
unchanged.”
Frequency Distribution: LINE CHART
 It is a way of summarizing data by showing the It is a way of presenting data by connecting data points
number of observations that belong in the with line segments. Line charts are usually used for
different categories or classes (also called group temporal data (data sets with time as one of the
data). variables). More than one line chart may be drawn on a
 Frequency may be expressed as absolute single plot area, especially if comparisons are to be
frequency (actual frequency) or relative made.
frequency (normalized frequency; for each
class, relative frequency is equal to the absolute
frequency of the class divided by the total
absolute frequency for all classes).
 Frequency distribution may be presented in
graphical form as either a frequency histogram
(vertical bar chart), a frequency polygon (line
chart), or an ogive (line chart showing
cumulative frequency distribution).

BAR CHART
It is a way of presenting data by using either vertical
bars or columns (vertical/ column bar charts) or
horizontal bars (horizontal bar charts). Bar charts are
usually used for frequency histograms and categorical
data sets.

PIE CHART
It is a way of presenting data by utilizing the area of a
circle for easier comparison by dividing it into several
sectors depending on the relative frequencies. Pie
charts are usually used for categorical data.

PICTOGRAPH
It is a way of presenting data similar to a horizontal bar
chart but using symbols or pictures to represent the
magnitudes of data. Pictographs are usually used for
categorical data.

STATISTICAL MAP
It is a way of presenting data that makes use of maps
and colors or shades to represent magnitudes of data.
Statistical maps are used in represent magnitudes of
data. Statistical maps are used in representing
geographic data.
DOTPLOT
It is a way of representing data similar to a column bar
chart that makes use of dots to represent the data
magnitudes. Dotplots are used for numerical data with
small number of observations.
STEM-AND-LEAF DISPLAY
It is an effective and compact way to summarize raw
numerical data by using the first few digits as “stems”
and the last digits as “leaves”. Stem-and leaf displays
are used for numerical data with small to moderate
number of observations.
SCATTERPLOT
It is a way of presenting data by plotting the data as a
set of scattered points. It is used to determine
relationship between two variables for bivariate data
sets.
CONSTRUCTING A FREQUENCY DISTRIBUTION
Constructing a frequency distribution may be done with
the following step-by-step procedures:
1. Arrange the raw data into an array.
2. Divide the data into classes. It is preferred to
have equal interval sizes for each class. The
interval size may be selected but having too
small or too large of an interval size. If one has
no idea on what interval size should be used,
Sturges’ rule may be used, i.e., the
recommended number of classes K should be
equal to K=1+ 3.322log n
3. Determine the range of the data (equal to the
difference between the largest and the smallest
values of observation) and the determine the
interval size by dividing the range by the
number of intervals.
4. Make a table listing the class intervals, the class
marks (midpoint of each class) and the number
of observations falling within their respective
intervals. You may use either absolute or
relative frequencies.
 5. Plot the frequency histogram or frequency
polygon of the said frequency distribution.
6. If ogives are to be plotted, add another column
for cumulative frequency distributions (either
increasing/ less than ogive or
decreasing/greater than ogive).
SAMPLE PROBLEM
Consider the following data for the annual rainfall
intensity recorded in a certain place over a period of 29
years.
43.30 53.02 63.52 45.93 48.26 50.51 49.57 43.93 46.77
59.12 54.49 47.38 40.78 45.05 50.37 54.91 51.28 39.91
53.29 67.59 58.71 42.96 55.77 41.31 58.83 48.21 44.67
67.72 43.11
Make a frequency distribution and plot its frequency
histogram, frequency polygon, and ogives for the said
data.
1. Rearrange the raw data into an array. (This step
is skipped and left as an exercise for students.)
2. Divide the data into classes. Since there are
n=29 observations, the recommended number
of classes according to Sturges’ rule is equal to
K=1+ 3.222log n=1+3.222 log 29=5.858
3. Determine the class size. In this example, the
largest observation value is 67.72 while the
smallest observation value 39.91. Therefore,
the range is equal to 27.81. The class size is
27.81/6=4.635.
4. Make a table showing the frequency
distribution of the given data. The following
table is from the MS Excel worksheet made for
this example.

The plots for frequency histograms (top), frequency

polygons (middle), and ogives (bottom) are shown,
using absolute (left)and relative (right) frequencies. MS
Excel is also used for plotting.