What is Business Intelligence

What is Business Intelligence (BI)?

Business intelligence (BI) leverages software and services to transform data

into actionable insights that inform an organization’s business strategic and
tactical decisions.

What are BI tools?

BI tools access and analyze data sets and present analytical findings in
reports, summaries, dashboards, graphs, charts and maps to provide users
with detailed intelligence about the state of the business.

The term business intelligence often also refers to a range of tools that
provide quick, easy-to-digest access to insights about an organization's
current state, based on available data.

Is BI purely Descriptive?

No, BI not purely descriptive. It is not just about generating reports and
dashboards. You can see patterns and get some insights through
visualizations.

Is BI Prescriptive?

No. BI is not prescriptive. It does not tell what the decisions ought to be.

So what is the role of BI?

Although BI does not tell business users what to do or what will happen if
they take a certain course of action, so it is not prescriptive. BI is not solely
about generating reports or dashboards, so it is not just descriptive. Rather,
BI offers a way for people to explore or examine data visually or otherwise,
to understand trends and derive insights by streamlining the effort needed to
search for, merge and query the data necessary to make sound business
decisions.

Here is a good example of an overview of

BI: [Link]
[Link] (Links to an external site.)

Because data is at the foundation of BI, we will study the fundamentals of

data in the next few pages.

What is Data?
What is Data?

A piece of data is basically the value of some variable of interest. We will first
define a single unit of data, then a column of data, then a row of data and
then a data set. Lastly, we will distinguish between structured and
unstructured data.

A Single Unit of Data

A single unit of data is the value of a variable for an entity in the real world.
So a single unit of data represents the value of something in the real world.

An Example: Let's say we are interested in the number of days it rained last
month. Here, the entity is "last month" and the variable is "the number of
days it rained". The value of this variable for this entity is a single unit of
data. Suppose the value is 5. So, 5 is a single unit of data that describes the
number of days it rained last month.

Another Example: Let's say we are interested in the number of customers

waiting at a given checkout counter right now. So the entity is the given
checkout counter and the variable is the number of customers waiting right
now. Suppose its value is 4. So 4 is a single unit of data that describes the
number of customers waiting at a the checkout counter right now.

Another Example: Let's say we are interested in the hair color of the next
person at the checkout counter. Here the entity is the next person at the
checkout counter and the variable is the hair color. Suppose its value is
"Gray". So, Gray is a single unit of data that describes the hair color of the
next person at the checkout counter.

A Column of Data

When we have data for one variable for many entities, we get a column of
data.

An Example: if we have the "Number of Days it Rained" for all 12 months of

last year, we will have a column of 12 single units of data. The column of
data might look like this:

Example

Number of days it rained in

Month
the month

January 10
February 8

March 5

April 7

May 3

June 7

July 8

August 10

Septemb
12
er

October 9

Novemb
6
er

Decembe
5
r

Note that in the above table, only the second column is our data. The first
column is the identifier of the month for which the data appears in the
second column.

Another example: We can have a column of data for the "Number of

Customers Waiting Right Now" for all five checkout counters, as follows:

Example

Checkout Number of customers waiting

Counter right now

A 5

B 4

C 3

D 4
E 6

Again, note that in the above table, only the second column is the data. The
first column is the identifier of the checkout counter for which the data
appears in the second column.

Another Example: Suppose we collect data on the hair color for the next
several persons at the checkout counter. We may get the following column
of data:

Example

Pers Hair
on Color

Next1 Gray

Next2 Blonde

Next3 Red

Next4 Black

Next5 Brown

Next6 Blonde

Again, note that in the above table, only the second column is our data. The
first column is the identifier of the next person at the checkout counter for
which the data appears in the second column.

A Row of Data:

When we have data for several variables for a single entity, we get a row of
data.

For example, for the last month (our entity), we may have data for several
variables, such as # of days in the month, # of days it rained, the total
amount of rainfall in inches, # of days the high temperature crossed 80 deg
F.

The row of data for the last month might look like this:

Example

# of days in # of days it Total amount # of days the high

 of rainfall temperature crossed 80 deg
month rained
(in inches) F

31 10 8.6 2

Note that in the above table, only the second row is the row of data. The
first row contains the names of the variables.

A Data Set:

When you have data for multiple variables for multiple entities, we get a
matrix of data or a table of data or a data set.

For Example: if we have data for the above four variables for all 12 months
of last year, we might get a data set as follows:

Example

Total amount # of days the high

# of days # of days
Month of rainfall temperature crossed 80
in month it rained
(in inches) deg F

January 31 10 8.6 2

February 28 8 7.2 5

... ... ... ... ...

Novemb
30 6 3.4 6
er

Decemb
31 5 5.6 4
er

Note that in the above data sets, we have four variables (columns) of data.
The first column in the above table is the ID column that identifies the row or
the entity. We have 12 rows of data, one for each month. The above table
only shows four rows; the remaining 8 are hidden. In the above table, there
are 12 entities and four variables. Columns are also sometimes referred to
as "Fields". Entities appear in rows.

Technically, even one column of data or just one row of data is a special case
of a data set, although not a very interesting one. A "typical" Data Set will
have multiple columns and multiple rows.
By the way, the above definition of data is a traditional definition.
Traditionally all data appeared in rows and columns because all data was
structured data. But lately, we have what is called unstructured data. So,
these days, for example, a customer review or a comment posted on a web
site is also considered data. A picture posted by someone on Facebook is also
considered data. An audio file of a song or a speech posted on a web site is
also considered data. A video posted on YouTube is also data. A customer
review represents someone's thoughts; an image represents something in
the real world, etc. therefore they are data.

Although there is a lot of unstructured data in the world, there is also a lot of
structured data. It is very important to understand "Structured" data because
it is very rich in content and allows easier analysis.

Review of this lesson:

1. Single unit of data: value of one variable for one entity

2. Column of data: values of one variable for multiple entities

3. Row of data: values of multiple variables for one entity

4. Data Set: values of multiple variables for multiple entities

5. Structured vs. Unstructured data: The Data set as defined above is

for structured data. Besides such structured data, we have
unstructured data in the form of text, images, audio and video files.

Quantitative vs. Qualitative Variables/Data

Quantitative vs. Qualitative Data

Recall that data is essentially the value of some variable for some entity. If
the variable is quantitative in nature, its values will be numeric. Similarly, if
the variable is qualitative in nature, its values will be non-numeric (although,
sometimes, numbers may act as labels).

Examples of Quantitative Variables

1. The number of students in a class

2. The number of customers who visit an online retail website on a given

day

3. The amount of rainfall in a given month

4. The weight of a given patient

 5. The age of a given customer

The above variables are all quantitative in nature, because their values are a
result of a measurement, or a count or a calculation or an estimate.
Therefore, their values will be numeric. For example, there might be 24
students in a class (count); 85 customers might visit an online retail website
on a given day (count); there might be 8.6" of rainfall in a given month
(measured); the weight of a given patient might be 182 lbs (measured) and
the age of a given customer might be 48 (estimated or calculated). In all five
examples, the values, i.e. the data, are numeric.

Examples of Qualitative Variables

1. The hair color of hair of a customer who walks into a hair salon

2. The religion of a given person

3. The race of a given presidential candidate

4. The grade received by a student

5. The degree of customer satisfaction of a customer

The above variables are qualitative in nature because their values are not a
result of a measurement or a count or a calculation or an estimate. They are
either observed or assigned or specified or somehow determined and
expressed as a non-numerical value. For example the hair color of a
customer who walks in a hair salon might be "Blonde" (observed); the
religion of a given person might be "Christianity" (specified by the person);
the race of a given presidential candidate might be "White" (determined or
observed); the grade received by a student might be a "B+" (assigned) and
the degree of satisfaction of a customer might be "Excellent" (specified).
Sometimes, even qualitative variables may have a numerical value without
any numeric quality.

Caution: Sometimes, as mentioned above, the value of a qualitative

variable may be expressed as a number but without its numerical quality.
This situation can be confusing. For example, the T-Shirt on a Team player
might have a number such as 15. Although 15 is a number, the underlying
variable is T-Shirt number, whose value is not a result of measurement or a
count. It is simply specified or assigned. So, in this example, 15 is qualitative
data, even though it is numeric, because in the context of a T-shirt number, it
has no intrinsic numeric quality. For example, it is not considered greater
than a T-Shirt number of 10 or less than a T-Shirt number of 20. The number
merely serves as a label, with no intrinsic arithmetic value. It could just as
well be a letter from A to Z.

So, when you see a numeric data, most likely it is quantitative, but not
necessarily, so we have to be careful how we process such data. But if you
see a non-numeric data, it is definitely qualitative.

Review of this lesson

A given variable might be either quantitative or qualitative in nature.

If its value is the result of a measurement or a count, the variable is

quantitative in nature and the data corresponding to such a variable would
be numeric.

If its value is not measured or counted but rather observed or specified in

qualitative terms, the variable is qualitative in nature and the data
corresponding to such a variable would be non-numeric, although in some
cases, the value could be numeric, with no intrinsic numeric value.

Discrete vs. Continuous Variables/Data

Discrete vs. Continuous Variables/Data

A quantitative variable can be further classified as either discrete or

continuous. A discrete variable can only have certain values, with distinct
breaks between values. A continuous variable can assume any value in a
range of values. By any value, we mean having no breaks. Continuous
variables will have decimals. Theoretically, infinite number of decimal points
are possible, although in practice only a few decimal points are meaningful.
If we use say only two decimal points, then one could argue that there is a
break between two possible values, and hence the variable is discrete. For
example, the volume of water in a cup may be 10.55 ml or 10.56 ml. One
could argue that there is a break between these two consecutive values and
hence volume must be a discrete variable. However, it should be realized
that the break is only due to the problem of measuring the volume to more
decimal places. Theoretically, if there was a volume measuring device that
could measure the volume to infinite decimal places, then there would be no
breaks. Intrinsically, volume is a continuous variable, hence its value is
considered continuous data, even if it has breaks due to limited decimal
points.

Examples of discrete quantitative variables

 1. Number of students in a class (Possible values are any integer from 0
to say 100, assuming 100 is the max limit. But there are no possible
values of this variable between two integers.)

2. Number of days in a month that it rains (Possible values are any

integer from 0 to 31. But there are no possible values between two
integers.)

3. Number of political parties contesting an election (Possible values can

range anywhere from 2 to several, depending on the country).

Examples of continuous quantitative variables

1. Volume of a room (The value may have any number of decimal points.
No breaks are necessary between two values).

2. Height of a person, without rounding (When not rounding, height can

be measured up to any number of decimal points, depending on the
measuring device).

3. Distance between two random points on a paper.(distance can be

measured up to any number of decimal points depending on the
measuring device).

Caution: In practice, things can only be measured up to a small number of

decimal points. For example money is measured up to 2 decimal places
(pennies). So in practice, there are breaks in values when representing
money. Yet, money is considered a continuous variable. Similarly lengths or
heights or weights might be measured up to only one or two decimal points.
So, in practice, all variables are discrete. But theoretically, variables whose
values are a result of counting are considered discrete, whereas variables
whose values are a result of measurement, such as length, height, area,
volume and time are considered continuous.

Note: All qualitative variables are necessarily discrete by nature because

they assume only a certain number of possible values.

Review of this lesson

A quantitative variable is considered discrete if it can assume only a certain

number of values with distinct breaks between two consecutive values, such
as 1,2,3...n. A quantitative variable is considered continuous if it can assume
any value (i.e. up to any number decimal points), with no distinct breaks. In
practice, due to limitations of measurement, all variables are discrete
because there are always breaks, however small, but theoretically,
quantitative variables that are measured are continuous in nature, whereas
quantitative variables that are counted are discrete in nature.

Nominal vs. Ordinal Scale Variables/Data

Nominal vs. Ordinal Scale Variables

When a variable is Qualitative in nature, it can be further classified as

either having either a Nominal scale or an Ordinal scale. The values of a
nominal scale qualitative variable cannot be ordered from low to high. The
values of an ordinal scale qualitative variable can be ordered from low to
high.

Examples of Nominal Scale Qualitative Variables

 Race (e.g. White, Asian, Black, Hispanic : Cannot be ordered)

 Religion (e.g. Christianity, Judaism, Islam : Cannot be ordered)

 Color (e.g. Red, White, Green: Cannot be ordered)

 T-Shirt Number (e.g. 15, 10, 20, 9, 13: Can be ordered, but the
ordering is meaningless)

Examples of Ordinal Scale Qualitative Variables

 Customer Satisfaction (e.g. Poor, Fair, Good, Excellent: Can be

ordered)

 Grade (e.g. A, B, C,D,F: Can be ordered)

 Rank in college (e.g. Freshman, Sophomore, Junior, Senior: Can be

ordered)

Review of this lesson

A qualitative variable can be measured on either a nominal scale (ordering of

values is either not possible or is meaningless) or an ordinal scale (ordering
of values is possible and is meaningful).

A Tip: The way I remember which type of variable is nominal and which
ordinal is that "ord"inal variables can be "ord"ered (get it?)

Interval vs. Ratio Scale Variables/Data

Interval vs. Ratio Scale Variables/Data

Quantitative variables, whether discrete or continuous, can be measured on

either the Interval scale or the Ratio scale.
A variable measured on Interval scale has the following characteristics:
It has no natural zero
The ratio of two values is meaningless

A variable measured on Ratio scale has the following characteristics:

It has a natural zero
The ratio of two values is meaningful

Examples of variables measured on Interval Scale:

1. Temperature in Fahrenheit.
Why is it on Interval scale? Since 0 deg F does not imply no heat, there
is no natural zero. The ratio of 80 and 40 is 2. But that does not mean
a temperature of 80 deg F feels twice as hot as a temperature of 40
deg F. So the ratio is meaningless.

2. SAT score.
Why is it on Interval scale? An SAT score does not even have a value of
0; the minimum score is 200. The ratio of a score of 1200 and 600 is
2. But that does not mean that the student scoring 1200 is twice as
smart as the one scoring 600. So the ratio is meaningless.

Examples of variables measured on Ratio Scale:

1. Distance between two points.

Why is it on Ratio scale? If two points coincide, the distance is zero.
The ratio of 20 and 10 is 2 and it implies that the distance of 20 is
twice as large as the distance of 10. So the ratio is meaningful.

2. Sale Price.
Why is it on Ratio scale? A sale price of 0 implies free. The ratio of 30
to 10 is 3 and it implies that the item priced at $30 is three times more
expensive than the item priced at $10. So the ratio is meaningful.

Review of this lesson:

A quantitative variable measured on an Interval scale has no natural zero

and the ratio of two values is meaningless. A quantitative variable measured
on Ratio scale has a natural zero and the ratio of two values is meaningful.

Ungrouped vs. Grouped Data

Ungrouped vs Grouped Data

Sometimes data just appears without any grouping.
For example data in a column may look like this:

Example

Number of days it rained in

Month
the month

January 10

February 8

March 5

April 7

May 3

June 7

July 8

August 10

Septemb
12
er

October 9

Novemb
6
er

Decembe
5
r

The numeric data in the above table is ungrouped. To convert it into grouped
data, we will form a frequency table with groups such as:

Example

Frequency
# of days of
(# of
rain
months)

0-4 1
5-8 7

9 - 12 4

Total 12

The first column in the above table has the groupings. The second column
has frequencies. The groups are also called bins.

Grouping the data in this manner allows us to draw charts for visualizations.
For example, for the above grouped data, we can draw a chart as follows:

Note that such charting would not be have been possible without grouping
the data.

Population vs. Sample Data

Population vs Sample Data

When you have data for all the entities of a population, it is called population
data. When you have data for only a subset or a sample of entities of a
population, we have sample data. For example, if our population is defined
as all the buildings in a downtown or a particular city and if we are collecting
data on the heights of buildings, then population data would have the height
of every building in the downtown of this city, whereas sample data would
contain heights of only a subset of buildings.
The term "population" in data science, has a different meaning than the
general term "population". The general term refers to people. But in data
science, population is basically all entities in a defined scope. An entity need
not be a living thing, as in the example above.

Why did we learn so much about different types of data?

It is important to distinguish between different types of data because not

every visualization chart applies to every type of data. For example, while
we can draw a histogram for an ordinal variable or a grouped quantitative
variable and interpret the shape of the histogram, it makes no sense to draw
a histogram for a nominal variable and interpret its shape. Also, we cannot
draw a histogram of ungrouped data. It makes no sense to find averages of a
numeric data that happens to be on nominal scale. These are just some
examples of things that can go wrong. We will learn more about them in the
future modules. Unless we correctly identify the type of data and unless we
know what type of visualization chart is suitable to what type of data, we
may commit errors and we will not be very effective in gaining insights from
a given dataset.