CHAPTER 2

DESCRIPTIVE STATISTICS: NUMERICAL MEASURES

In the worlds of finance and accounting, numbers are more than just figures on a
page; they are the pulse of a business. However, raw data can be overwhelming.
Descriptive statistics provides the tools to organize, summarize, and visualize this data,
turning a chaotic spreadsheet into a clear narrative of performance and risk.

Chapter Learning Outcomes

At the end of the chapter, the student will be able to:

1. Identify and know the use of descriptive statistics in business, finance and
accounting.
2. Analyzes data to get the measures of central locations and assess data variability
3. Utilize Microsoft Excel to automate statistical analysis.

Introduction

While we previously explored how to present data using tables and graphs, this
chapter shifts the focus to numerical descriptive statistics. These measures offer a
precise, quantitative way to summarize information beyond visual representation alone.

We will examine four primary categories of numerical measures in terms of:

a. Location - Pinpointing where the data centers.

b. Dispersion - Measuring how spread out the values are.
c. Shape - Identifying the distribution's symmetry or "peak."
d. Association - Determining how different variables relate to one another.

It is also important to distinguish between the source of your data. When these
values are calculated from a sample, they are referred to as sample statistics; however,
when they represent an entire population, they are known as population parameters.
This chapter also introduces a variety of Excel statistical functions designed to perform
descriptive analysis. These tools allow you to efficiently calculate numerical summaries
and gain deeper insights into your data directly within the spreadsheet environment.
Lesson 1. Measures of Location
In descriptive statistics, Measures of Location (often called measures of central
tendency or relative standing) are numerical values used to identify the center, typicality,
or specific position of a data point within a distribution. Instead of looking at a raw list of
thousands of numbers, these measures allow a business leader to grasp the "heart" of
the data in a single glance. The mean, weighted mean, median and mode are measures
of central tendency, which helps identify the middle or most representative value of a data
set. While the percentiles and quartiles measure the relative standing or position, where
a specific value falls in relation to the rest of the group.

MEAN

The mean, often referred to as the arithmetic average, is the most fundamental
measure of central location. It represents the "balance point" of a dataset, calculated by
summing all the individual values and dividing that total by the number of observations.
Because it incorporates every piece of data in the set, it is highly sensitive; even a single
extremely high or low value (an outlier) can pull the mean toward it, potentially
misrepresenting the "typical" experience of the group. Additionally, mean is a good
measure for interval and ratio data, as well as, inferential statistics is based on
mathematical properties of the mean. If the data is taken from a sample mean is denoted
by 𝒙
̅. If the data is taken from a population, mean is denoted by 𝝁 (mu).

SAMPLE MEAN POPULATION MEAN

𝜮𝒙𝒊 𝒙𝟏 +𝒙𝟐 + …+𝒙𝒏 𝜮𝒙 𝒙𝟏 +𝒙𝟐 + …+𝒙𝑵
̅=
𝒙 = 𝝁 = 𝑵𝒊 =
𝒏 𝒏 𝑵

Example 1. The college placement office would like to get the data and send
questionnaire to a sample of business school requesting information on monthly starting
salaries. The table below shows the collected data.

Table 1. Monthly starting salaries for a sample of 12 business school graduates

Monthly Starting Salary Monthly Starting
Graduate Graduate
(PhP) Salary (PhP)
1 25,000.00 7 28,500.00
2 28,000.00 8 26,500.00
3 25,600.00 9 27,800.00
4 26,500.00 10 32,000.00
5 30,000.00 11 30,500.00
6 35,000.00 12 26,500.00
 The mean monthly starting salary for the sample of 12 business college graduates
(n = 12) is computed below.

𝜮𝒙𝒊 x1 +x2 + …+x12

̅=
𝒙 =
𝒏 12

25,000+⋯+26,500
= 12

341,900.00
= = 𝟐𝟖, 𝟒𝟗𝟏. 𝟔𝟕
12

In business, the mean is more than just a mathematical average; it serves as the
primary benchmark for performance and risk assessment. It provides a single, simplified
value that allows professionals to compare current results against historical trends or
industry standards. In finance, the mean is the foundation of Expected Return. Investors
use the arithmetic mean of historical stock returns to estimate what an asset might earn
in the future.

While the mean is powerful, it can be "fooled" by outliers. For example, if a

business has ten small clients and one massive corporate contract, the mean revenue
might look great, but it doesn't reflect the reality of the ten smaller struggling accounts.

WEIGHTED MEAN

In many business and finance situations, not all data points are created equal. A
Weighted Mean is used when certain values in a dataset are more important—or have a
greater "weight"—than others. If you use a mean, you treat every item as if it has the
same impact, which can lead to misleading conclusions.

Consider these common situations where weighted mean is used.

Investment portfolios. If you own ₱ 90,000 of Stock A (which gained 10%) and ₱10,000 of
Stock B (which lost 50%), your "average" return isn't the simple average of +10 and -50. Since
you have much more money in Stock A, its performance "weighs" more heavily on your total
wealth.

Inventory Accounting. In the Philippines, many businesses use the WAC method for tax and
reporting purposes (under PFRS). Since prices for raw materials like fuel, sugar, or flour
fluctuate constantly, you cannot simply pick the most recent price to value your entire
warehouse. The Weighted Mean ensures that the "cost of goods sold" reflects the actual
average price paid for every single unit currently in stock, providing a much more accurate
picture of profit margins.

To calculate the weighted mean (𝒙 ̅𝒘 ) you multiply each value (𝑥𝑖 ) by its
corresponding weight (𝑤𝑖 ), sum those products, and then divide by the total of all weights.
 Σ(𝑤𝑖 𝑥𝑖 )
̅𝒘 =
𝒙
Σ𝑤𝑖

Example 2. A local retailer sourcing sacks of rice at different prices over a month.
Batch Quantity (Weight w) Cost per Unit (x) Total Cost (w*x)
Batch 1 100 sacks ₱2,100 ₱210,000
Batch 2 200 sacks ₱2,050 ₱410,000
Batch 3 50 sacks ₱2,250 ₱112,500
Totals 350 sacks ₱732,500

Σ(𝑤𝑖 𝑥𝑖 ) 210,000 + 410,00 + 112,000

̅𝒘 =
𝒙 = = ₱ 𝟐, 𝟎𝟗𝟐. 𝟖𝟔
Σ𝑤𝑖 350
If the business owner used the simple average (₱2,133.33) to set their selling
price, they might overprice their goods and lose customers to competitors. Because they
bought the bulk of their stock (200 sacks) at the cheaper price of ₱2,050, the weighted
mean (₱2,092.86) gives them a much more accurate "break-even" point.

In accounting, this is known as the Weighted Average Cost (WAC) method, and
it is a standard practice for managing inventory under the Philippine Financial Reporting
Standards (PFRS).

MEDIAN

The Median is another measure of central location for a data. The median is the
value in the middle when the data are arranged in ascending order (smallest value to
largest value). With an odd number of observations, the median is the middle value. If the
number of observations is even, the median is the average of the two middle values.

When we talk about "average" in the Philippines business setting—especially

regarding household income or real estate prices—the Median is often a much more
honest measure than the Mean. While the Mean is a "balance point," the Median is the
exact middle point of your data.

Example 3. Supposedly you are looking at the monthly salaries of five employees (n = 5)
in small start-up business in Manila.

Monthly Salary
Juan dela Cruz ₱ 20,000
Maria Basilia ₱ 25,000
Pedro Penduko ₱ 30,000
Perla Santiago ₱ 35,000
Pablo Nase (CEO) ₱ 500,000
 If you calculate the Mean, the average salary looks like ₱122,000. This is
misleading! No one actually earns anywhere near that except the CEO. However, the
Median is the middle value: ₱30,000. This gives a much more "typical" picture of what an
employee earns (because n = 5 is odd, the median is the middle value).

Example 4. By using the data in Table 1. We arrange first the values of the starting salary
in ascending order, where n = 12, an even number

25,000 25,600 26,500 26,500 26,500 27,800

28,000 28,500 30,000 30,500 32,000 35,000

27,800 + 28,000
𝑀ⅇ𝑑𝑖𝑎𝑛 = = 𝟐𝟕, 𝟗𝟎𝟎
2

The median is the measure of location most often reported for annual income and
property value data because a few extremely large incomes or property values can inflate
the mean. The median is not influenced by extreme values or skewed distribution, and for
ordinal data, median is more appropriate than mean. In such cases, the median is
preferred measure of central location.

MODE

The third measure of location is the mode. Mode is the value that occurs with
greatest frequency. To illustrate the identification of the mode, consider example 1 with
the monthly starting salaries for business school graduates.

25,000 25,600 26,500 26,500 26,500 27,800

28,000 28,500 30,000 30,500 32,000 35,000

The only monthly that occurs more than once is ₱ 26,500, because it has the
greatest frequency, it is the mode.

There are situations, in which the greatest frequency occurs at two or more
different values. If the data contain exactly two modes, that data are bimodal. If the data
contains more than two modes, that data is multimodal. In multimodal cases, the mode
is almost never reported because listing three or more modes would not be particularly
helpful in describing a location for the data. Additionally, the mode is important measure
of location for qualitative data (nominal and ordinal).

Example 5. The data is from a sample of 50 soft drink purchases.

Coke Classic Sprite Pepsi-Cola

Diet Coke Coke Classic Coke Classic
Pepsi-Cola Diet Coke Coke Classic
Diet Coke Coke Classic Coke Classic
Coke Classic Diet Coke Pepsi-Cola
Coke Classic Coke Classic Royal
Royal Sprite Coke Classic
Diet Coke Pepsi-Cola Diet Coke
Pepsi-Cola Coke Classic Pepsi-Cola
Pepsi-Cola Coke Classic Pepsi-Cola
Coke Classic Coke Classic Pepsi-Cola
Royal Pepsi-Cola Pepsi-Cola
Sprite Coke Classic Coke Classic
Coke Classic Sprite Royal
Diet Coke Royal Pepsi-Cola
Coke Classic Pepsi-Cola Sprite
Coke Classic Diet Coke

By constructing the frequency distribution table of soft drink purchase, the table is
shown below:

Table 2. Frequency distribution of soft drink purchases

Soft Drink Frequency
Coke Classic 19
Diet Coke 8
Pepsi-Cola 13
Royal 5
Sprite 5
Total 50

By looking at the table, the mode, the most frequently purchased soft drink is Coke
Classic. For this type of data, it makes no sense to speak of the mean and median, the
mode is providing the most information of interest, the most frequently purchased soft
drink.

The use of Mode as measure of central location in business are used in inventory
and retail management. Example, the mode is the ultimate tool for Stock Keeping Units
(SKUs). If a shoe store sells sizes 5 to 12, the mean size might be 8.4—but you can't
manufacture a size 8.4 shoe. For marketing and consumer preferences, in conducting
surveys (e.g. "Which milk tea flavor do you prefer?"), you cannot calculate a "mean" or
"median" for flavors like Wintermelon or Taro because they aren't numbers. If
"Wintermelon" appears most often in the survey results, it is the mode, and the marketing
team will focus their budget on that flavor.
USING EXCEL TO COMPUTE MEAN, MEDIAN AND MODE

Excel provides functions for computing the mean, median and mode. We will
illustrate the use of these Excel functions by computing the mean, median and mode for
the monthly starting salary data in Table 1.

Figure 1. Excel worksheet used to compute mean, median and mode for starting salaries

Enter Data: Labels and the starting salary data are entered into cells A1:B13 of the
worksheet
Enter Functions and Formulas: Excel’s AVERAGE function can be used to compute
the mean by entering the formula to cell E1:
=AVERAGE(B2:B13)
Similarly, for the median and mode, the formulas are =MEDIAN(B2:B13) and
=MODE(B2:B13) are entered into cells of E2 and E3. The labels Mean, Median and
Mode are entered into cells D1, D2 and D3 or D1:D3.
PERCENTILES

A percentile provides information about how the data are spread over the interval
from the smallest value to the largest value. For data that do not contain numerous repeat
values, the 𝑝th percentile divides the into two parts. Approximately p percent of the
observations have values less than the 𝑝th percentile; approximately (100 – 𝑝) percent
of the observations have values greater than the 𝑝th percentile. The 𝑝th percentile is
defined as follows.

PERCENTILE

The 𝑝th percentile is a value such that at least 𝑝 percent of the observations are less
than or equal to this value and at least (100 – 𝑝) percent of the observations are greater
than or equal to this value.

While the mean gives you an average and median gives you the middle value,
percentiles help you understand rank and distribution, which is often more useful for
setting benchmarks or identifying top and bottom performers. In the aspect of human
resource, many companies in the Philippines use "Market Percentiles" to stay
competitive. A firm might decide to pay its software engineers at the 75th percentile of the
industry average. This suggests that they pay more than 75% of other companies, helping
them attract top-tier talent. For customer analytics, marketers use percentiles to segment
customers by spending. The "Top 10%" of spenders (those in the 90th percentile and
above) might be invited to a VIP loyalty program.

The following procedure can be used to compute the 𝑝th percentile.

CALCULATING THE 𝒑TH PERCENTILE

Step 1. Arrange the data in ascending order (smallest value to largest value.
Step 2. Compute an index 𝑖
𝑝
𝑖= ൬ ൰𝑛
100

Where 𝑝 is the percentile of interest and 𝑛 is the number of observations.

Step 3. (a) If 𝑖 is not an integer, round up. The next integer greater than 𝑖 denotes
the position of the 𝑝th percentile.
(b) If 𝑖 is an integer, the 𝑝th percentile is the average of the values in the
position 𝑖 and 𝑖 + 1
Example 6. We are going to use the data in Table 1, regarding the starting salary. We are
going to determine the 85th percentile of the salary.
Step 1. Arrange the data in ascending order, where n = 12

25,000 25,600 26,500 26,500 26,500 27,800 28,000 28,500 30,000 30,500 32,000 35,000

Step 2. Compute the index (i)

𝑝 85
𝑖= ( )𝑛 = ൬ ൰ 12 = 10.2
100 100
Step 3. Because 𝑖 is not an integer, round up. The position of the 85th percentile is the
next integer greater than 10.2, the 11th position.
By looking at the data, we see the 85th percentile is the data value in the 11th
position 32,000

Example 7. Let us consider the 50th percentile for the starting salary, by applying step 2.
𝑝 50
𝑖= ( )𝑛 = ൬ ൰ 12 = 6
100 100
6 is an integer, step 3 (b) states that the 50th percentile is the average of the sixth
and the seventh data values, thus the 50th percentile is
27,800 + 28,000
= 27,900
2

The 50th percentile is 27,900, and you have to take note that the 50th percentile is
also the median.

QUARTILES

Quartiles are a specific type of percentile that divide a sorted dataset into four
equal parts, each representing 25% of the total observations. They are essential in
business because they provide a clearer picture of data distribution than a simple average
can, helping managers identify "middle-of-the-pack" performance versus "top-tier" results.

The Three Key Quartiles

1. First Quartile (Q1). The 25th percentile. It separates the lowest 25% of data from the
rest. In business, this often represents "budget" or "underperforming" categories.
2. Second Quartile (Q2). The 50th percentile, also known as the Median. it splits the
data exactly in half.
3. Third Quartile (Q3). The 75th percentile. It separates the top 25% of data. This usually
represents "premium" or "high-performing" categories.
 In the Philippines, companies like BPO firms or Banks use quartiles to manage
their payroll budgets through a "Salary Structure." Imagine a company analyzing the
monthly salaries of 100 Customer Service Representatives:
• Q1 (25th Percentile) = ₱18,000. New hires or trainees usually fall here.
• Q2 (Median) = ₱25,000. This is the standard "market rate" for a typical employee.
• Q3 (75th Percentile) = ₱35,000. This is reserved for "High Performers" or senior-
level staff.

In terms of budgeting, if a manager sees that 60% of their staff are paid in the 4th
Quartile (above Q3), the department is likely over-budget and needs to justify the high
labor costs. Now for retention of employees, if the company’s Q3 (₱35,000) is lower than
a competitor's Q1, they are at high risk of losing their best employees to other firms.

In computing for quartiles, we can use the formula in computing percentiles. For
illustration, we are going to use the data of starting salary that have been sorted out.

Example 8. Starting salary of 12 college graduates in business school

25,000 25,600 26,500 26,500 26,500 27,800 28,000 28,500 30,000 30,500 32,000 35,000

The computation of quartiles Q1 and Q3 requires the use of the rule for finding the
25th and 75th percentiles. The solution is as follows:

For Q1
𝑝 25
𝑖= ( )𝑛 = ൬ ൰ 12 = 3
100 100

For Q3
𝑝 75
𝑖= ( )𝑛 = ൬ ൰ 12 = 9
100 100

Because the i is an integer for Q1 and Q3, we will average the third and fourth data
values for Q1, and we will get the average of the ninth and tenth value for Q3.
26,500+26,500 30,000+30,500
= 26,500 = 30,250
2 2

The quartiles divide the starting salary data into four parts, with each part
containing the 25% of the observation.

25,000 25,600 26,500 | 26,500 26,500 27,800 | 28,000 28,500 30,000 | 30,500 32,000 35,000
Q1 = 26,500 Q2 = 27,900 Q3 = 30,250
USING EXCEL TO COMPUTE PERCENTILE AND QUARTILES

Computing the percentile and quartiles in Excel is easy. Excel has PERCENTILE
and QUARTILE functions than can be used in getting the percentile and quartile of the
starting salary data.

In Excel, the function is =PERCENTILE(array, k) where the array is the range of

data that defines relative standing, in excel these are the data from B2 to B13, expressed
in excel B2:B13. The k is the kth percentile of values in a range, example if your want to
get the 85% percentile which refers the k, it is expressed in decimal point 0.85.

The quartile has quartile functions =QUARTILE(array, quart). The array is the
range of data similar with percentile, and the quart have 4 choices, 0 (minimum value), 1
(1st quartile), 2 (Median value or 2nd quartile), 3 (3rd quartile) and 4 (maximum value). This
is illustrated in Figure 2, in getting the percentiles and quartiles of the starting salary.

Figure 2. The Excel functions of Percentiles and Quartiles

Lesson 2. Measures of Variability
In addition to measures of location, measures of variability are often desirable
consider in a dataset, because it will tell us how spread our data points is from the center.
In business, "average" is rarely the full story. To illustrate, two delivery companies might
both have a mean delivery time of 3 days. However, if Company A delivers everything in
exactly 3 days, and Company B delivers some in 1 day and others in 5 days, Company
B has high variability. In business, high variability usually means unpredictability and risk.
In finance, variability is the literal definition of risk. The higher the standard deviation of a
stock’s price, the riskier the asset is considered. For customers satisfaction, customers
value consistency. If a Jollibee branch serves a burger that tastes different every time you
visit (high variability), customer loyalty will drop, even if the "average" taste is okay.

In this lesson, we will discuss the different measure of variability or dispersion. The
measures of variability include Range (R), Interquartile Range (IQR), Variance and
Standard Deviation, and Coefficient of Variation (CV).

RANGE

The simplest measure of variability is the range. To get the Range (R) of the data,
we get the difference between the largest value and smallest value in a data set.

RANGE
𝑅𝑎𝑛𝑔𝑒 = 𝐿𝑎𝑟𝑔𝑒𝑠𝑡 𝑣𝑎𝑙𝑢𝑒 – 𝑆𝑚𝑎𝑙𝑙𝑒𝑠𝑡 𝑣𝑎𝑙𝑢𝑒

By using the data in Table 1, about the starting salary, from the data we get the
largest value of the data set which is 35,000 and the smallest value is 25,000. To get the
range of the data set, we get the difference of these two values, 35,000 – 25,000 =
10,000.

The range is the easiest to measure variability, but it is seldom used as the only
measure. The range is sensitive to outliers, which its greatest flaw. Because you only use
the two most extreme values, this unusual data point makes the dataset to look more
spread than it actually is. To illustrate this flaw, if nine employees earn ₱25,000 and one
executive earns ₱1,000,000, the range is ₱975,000. This suggests massive variation,
even though 90% of the staff have identical pay.

The second thing is it ignores the "internal" data. The range tells us nothing about
how the data is distributed between the two endpoints. Two datasets can have the exact
same range but look completely different.
Set A: 10, 50, 50, 50, 90 (most of the values are in the middle)
Set B: 10, 11, 12, 89, 90 (most of the values are at the extreme endpoints)

Both have Range = 80 but their internal behavior is opposite.

INTERQUARTILE RANGE

The Interquartile Range (IQR) is a measure of variability that focuses on the

"middle half" of a dataset. It is calculated by subtracting the First Quartile (Q1) from the
Third Quartile (Q3).

INTERQUARTILE RANGE
𝐼𝑄𝑅 = 𝑄3 – 𝑄1

While the standard Range looks at the distance between the absolute highest and
lowest points, the IQR looks at the spread of the middle 50% of the data. By ignoring the
top 25% and the bottom 25%, it effectively removes the "noise" caused by extreme values
as show in Figure 3.

Figure 3. Box Plot showing quartiles, interquartile range and outliers.

Using the data on the starting monthly salary in table 1 and the solve examples
of Example 8 for the values of 𝑄1 (25,500) and 𝑄3 (30,250) to get the value of the IQR.

𝐼𝑄𝑅 = 𝑄3 – 𝑄1 = 30,250 − 25,500 = 4,750

These means that the middle 50% of starting monthly salaries are clustered within
a range of ₱4,750, specifically falling between ₱25,500 and ₱30,250. This indicates that
the majority of entry-level earners in this dataset experience very similar financial starting
points.

In business and finance, the IQR is the "gold standard" for understanding the
typical spread of a process without being distracted by chances. The most common
technical use of IQR is the 1.5 x IQR Rule. Any data point that falls more than 1.5 times
the IQR above 𝑄3 or below 𝑄1 is statistically flagged as an outlier (see Figure 3). In a
Philippine bank, for example, if most withdrawals are between ₱2,000 and ₱10,000, a
sudden withdrawal of ₱1,000,000 would be flagged for review based on this logic.
 Moreover, the HR departments use IQR to evaluate employee productivity. If the
IQR of sales made by a team is small, it means most employees are performing
consistently. A large IQR suggests a massive gap between your "average" and "top"
performers. For inventory management, retailers like Mercury Drug, the IQR of daily
demand for a specific medicine helps them decide how much "buffer stock" to keep. They
want to be prepared for the middle 50% to 75% of likely scenarios, rather than the one-
off day someone bought the entire shelf.

However, interquartile range has its own setbacks. There is a loss of information
by discarding the upper and lower 25%, we are literally throwing away half of our data. In
some business cases, those extremes are actually the most important parts (e.g., your
highest-paying 5% of customers or your most critical system failures). Additionally, it is
sensitive to sample size. in very small datasets, 𝑄1 and 𝑄3 can shift dramatically with the
addition of just one or two new numbers, making the IQR fluctuate and appear unreliable.

VARIANCE

Variance is a more comprehensive measure of variability. It means it is a measure

of variability that utilizes all the data. It measures the average of the squared differences
between each data point and the arithmetic mean. In simpler terms, variance tells us how
"stretched" or "squeezed" the entire distribution is. If the variance is zero, all values in the
dataset are identical. As the variance increases, it indicates that the data points are
moving further away from the average.

If the data is from a population, variance is called population variance, and

denoted by a Greek symbol 𝝈𝟐 (sigma squared). The formula is as follows:

POPULATION VARIANCE
σ(𝑥𝑖 − 𝜇)2
𝜎2 =
𝑁

However, most statistical application, data being analyzed are from samples, and
it is called sample variance denoted by the symbol 𝒔𝟐 , and the denominator is 𝒏 − 𝟏.

SAMPLE VARIANCE
2
σ(𝑥𝑖 − 𝑥ҧ )2
𝑠 =
𝑛−1
Example 9. To demonstrate the computation of variance, we will use the data of five
college classes.

Squared Deviation
Deviation from
Number of Students Mean of Class from about the
about the Mean
in a Class (𝒙𝒊 ) Size (𝒙
̅) Mean
(𝒙𝒊 − 𝒙
̅)
̅ )𝟐
(𝒙𝒊 − 𝒙
46 44 2 4
54 44 10 100
42 44 -2 4
46 44 2 4
32 44 -12 144
Total 0 256

σ(𝑥𝑖 − 𝑥ҧ )2 256
𝑠2 = = = 64
𝑛−1 4

In business, variance is the mathematical foundation for measuring uncertainty.

While the mean tells a manager what to expect "on average," variance tells them how
much they should prepare for things to go wrong (or exceptionally well). For Investment
Risk Analysis, in finance, variance is the mathematical representation of volatility. If a
stock has a high variance in its historical returns, it is considered a high-risk investment
because its future price is harder to predict. Moreover, in quality control of the product,
factory managers use variance to monitor machine precision. If the variance of a
component's thickness increases, it means the machine is becoming unstable and
producing defective parts.

However, variance as measure of variability has its own drawbacks. Because of

its squared units, the formula squares the differences (to ensure negative differences
don't cancel out positive ones), the result is in squared units. To illustrate this, If you are
measuring salaries in Philippine Pesos (₱), the variance is expressed in "Squared Pesos."
This is impossible to visualize or explain in a boardroom meeting. Additionally, it is
sensitive to outliers, variance is heavily influenced by extreme values. Because the
distance from the mean is squared, an outlier that is far away has a disproportionately
massive impact on the final result.

But variance is the "engine" behind most advanced business analytics. While it is
hard to explain because of the squared units, it is the essential first step to calculating
Standard Deviation and Beta (market risk).
STANDARD DEVIATION

Standard deviation is defined as positive square root of the variance. To

understand Standard Deviation, think of it as the "average distance" every data point sits
from the mean. While the mean tells you where the center is, the standard deviation tells
you how reliable that center actually is. Now, assuming you are managing BPO teams’
performance or tracking the price of rice in a local market, the standard deviation is your
primary tool for measuring consistency.

STANDARD DEVIATION
𝜎 = ඥ𝜎 2 (Population standard deviation)
𝑠 = ඥ𝑠 2 (Sample standard deviation)

Example 10. Recall, the sample variance of the class sized in five college classes in
Example 9, with 𝑠 2 = 64, by getting the square root of the variance, the standard deviation
is 𝑠 = ඥ64 = 8. This means that, on average, a typical class in this group deviates from
the mean by about 8 students.

The application of standard deviation in business are substantial. In case of quality

control, manufacturing companies (like Del Monte or Fantech) use standard deviation to
ensure products are identical. If a machine fills cans with 330ml of juice, a low standard
deviation means every can is nearly perfect. A high standard deviation means some cans
are overflowing while others are half-empty, leading to waste and customer complaints.
Furthermore, standard deviations are used in investment. On the Philippine Stock
Exchange (PSE), investors look at the standard deviation of a stock’s daily returns. Low
standard deviation means that the stock price is stable. While a high standard deviation,
the stocks change erratically (e.g., a speculative tech startup or mining firm). This
represents higher risk.
But there are also limitations of standard deviation, like variance, it is sensitive to
outliers. Because the formula involves squaring the distances from the mean, a single
extreme value (a "Black Swan" event) can drastically inflate the standard deviation,
making a stable process look volatile. Also, standard deviation is most meaningful when
the data follows a "Bell Curve” or normally distributed. If the data is heavily skewed (like
wealth distribution, where a few people have billions), the standard deviation can be
misleading. Additionally, it is difficult in comparing different things. For instance, a
standard deviation of ₱5,000 is "small" if you are selling cars, but it is "massive" if you are
selling t-shirts. To fix this, businesses use the Coefficient of Variation (CV).
COEFFICIENT OF VARIATION

The Coefficient of Variation (CV) is a relative measure of variability. While the

Standard Deviation tells you the "spread" in absolute terms (like Pesos or Centimeters),
the CV expresses that spread as a percentage of the mean.

COEFFCIENT OF VARIATION
𝑆𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝐷𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛
𝐶𝑉 = ቆ ቇ 𝑥 100%
𝑀𝑒𝑎𝑛

The main strength of the CV is that it allows you to compare the volatility or risk of
two datasets that have completely different scales or units.

Example11. When comparing investments with different prices, Stock A (Blue Chip with
a Mean price = ₱2,000; Standard Deviation = ₱100, Stock B (Penny Stock) with a Mean
price = ₱5; Standard Deviation = ₱2. If we only look at the Standard Deviation, Stock A
looks "riskier" because ₱100 is a bigger number than ₱2. But if we calculate the CV:

100
Stock A CV 𝐶𝑉 = (2000) 𝑥 100% = 5%

2
Stock B CV 𝐶𝑉 = (5) 𝑥 100% = 40%

We could see that Stock B is actually 8 times more volatile relative to its price than
Stock A.

Example 12. Using the data of the 5 class sizes of the school in example 9 and example.
We have a mean, 𝑥ҧ = 44 and standard deviation, 𝑠 = 8.
8
𝐶𝑉 = ൬ ൰ 𝑥 100% = 18.2%
44

The Coefficient of Variation shows that the variation in class sizes is relatively low,
representing less than one-fifth or 18.2% of the average class size. This suggests that
the school's enrollment is consistent and predictable, allowing for efficient classroom and
resource management.
USING EXCEL’S DESCRIPTIVE STATISTICS TOOL

Figure 4. Excell worksheet to compute mean, median, mode, variance and standard
deviation
LESSON 3. MEASURES OF DISTRIBUTION SHAPE,
RELATIVE LOCATION AND DETECTING OUTLIERS

In the world of business and finance, data isn't just a collection of numbers; it’s a
story about performance, risk, and efficiency. To read this story, we use measures of
Distribution Shape, Relative Location, and Outlier Detection. These tools allow data
analysts to move beyond simple averages and understand the "personality" of their
data—whether it’s predictable, skewed by high earners, or hiding errors that could lead
to costly decisions.

DISTRIBUTION SHAPE

The four panels in Figure 5 a clear visual comparison of how skewness—a

measure of the asymmetry of a probability distribution—affects the "tail" and the "peak"
of a dataset. In the business world, recognizing these patterns is the first step in identifying
risk and opportunity.

FIGURE 5. Histograms showing the skewness for four distributions

Panel A is moderately skewed to the left with skewness is -0.85. In this histogram,
the "tail" of the data stretches toward the lower values on the left. This represents a
situation where most observations are high, but a few significantly low values are pulling
the average down. Now, imagine Product Quality Scores where most units score near
90%, but a few defective units score 20% or 30%. In this case, the Mean will be lower
than the Median, and Median is lower than the Mode (Mean < Median < Mode).
 In Figure 5 we can see that Panel B is moderately skewed to the right with
skewness is 0.85, The tail stretches toward the higher values on the right. Most data
points are clustered at the lower end, with a few higher values "stretching" the distribution.
In this situation, Mean is greater than the Median, and Median is greater than the Mode.
(Mean > Median > Mode). In business, this skewness often represents Employee
Salaries. Most staff earn a base wage (the peak), while a few managers and executives
earn significantly more, creating that right-hand tail.

For Panel C, this is the classic "Bell Curve" (Normal Distribution) where the data is
perfectly balanced. The left side is a mirror image of the right. In a symmetric distribution,
the Mean, Median, and Mode are all equal (Mean = Median = Mode). This is common in
Manufacturing Tolerances, where a machine is calibrated to cut a part at exactly 50mm;
most parts are 50mm, with an equal (and small) number of parts being slightly too large
or slightly too small.

In the figure, Panel D is highly skewed to the right (Skewness = 1.62). This shows
a much more extreme version of Panel B. The peak is very sharp and located at the far
left, with a very long, thin tail extending far to the right. The vast majority of the data is
concentrated at the very bottom of the scale, but the "outliers" or extreme values are very
far away. This is typical for Insurance Claims or Start-up Returns. Most insurance
policies result in Php 0.00 claims, but a very tiny percentage (the far-right tail) results in
massive million-dollar payouts. In this scenario, the Mean is much higher than the Median,
which can be misleading if you only look at the average.

TABLE 10. COMPARISON OF KEY INDICATORS

Feature Left Skew (A) Symmetric (C) Right Skew (B/D)
Tail Direction Left None (Equal) Right
Mean vs Median
Mean < Median < Mode Mean = Median = Mode Mean > Median > Mode
vs Mode
Typical Example Life Expectancy Standardized Test Household Income
Scores
Note: Here is a tip, always "follow the tail." If the tail points to the right (the positive side
of the X-axis), it is a positive skew. If the tail points to the left (the negative side), it is a
negative skew.

Skewness

𝒏 ̅ 𝟑
𝒙𝒊 − 𝒙
𝑺𝒌𝒆𝒘𝒏𝒆𝒔𝒔 = ∑൬ ൰
(𝒏 − 𝟏)(𝒏 − 𝟐) 𝒔

In Excel, there is a function in getting the value of the skewness of your data. The
function is =SKEW (array of data). Using the starting salary data the function is
=SKEW(B2:B13) and press enter. The result is positive highly skewed value, 1.0884

TABLE 11. SUMMARY OF SKEWNESS NUMERICAL INDICATORS

Skewness Value Description Tail Direction
Between -0.5 and 0.5 Approximately Symmetric Symmetric (normal distributed
0.5 to 1.0 Moderately Skewed Right (positive)
-0.5 to -1.0 Moderately Skewed Left (Negative)
> 1.0 Highly Skewed Right (Positive)
< -1.0 Highly Skewed Left (Negative)
z-Scores

Relative Location refers to the position of a specific data value in relation to the
rest of the dataset. While central tendency (like the mean) tells us where the middle is,
and measures of variability (like range) tell us how spread out the data is, relative location
tells us exactly where an individual observation "stands" within that spread. The two
primary tools used to define relative location are Z-scores and Percentiles.

z-Score
̅
𝒙𝒊 − 𝒙
𝒛𝒊 =
𝒔
Where:
𝒛𝒊 = the z-score for 𝒙𝒊
̅ = the sample mean
𝒙
𝒔 = the sample standard deviation

The z-score is called the standardized value. The z-score can be interpreted as
the number of standard deviations of 𝑥𝑖 is from the mean (𝒙 ̅). For example, 𝑧1 = 1.2 would
indicate that 𝑥1 is 1.2 standard deviations greater than the sample mean.
Similarly, 𝑧2 = -0.5 would indicate that 𝑥2 is 0.5 or 1/2, standard deviation less than
the sample mean. A z-score greater than zero occurs for observations with a value greater
than the mean, and a z-score less than zero occurs for observations with a value less
than the mean. A z-score of zero indicates that the value of the observation is equal to
the mean.

Let us consider the data from the number of students per class in the previous
example. We compute the difference of the 𝒙𝒊 to the mean, we already computed the
value for the mean which is 44 and the standard deviation is 8. The z-score on the fifth
observation (-1.50) shows it is farthest from the mean, that it is 1.50 standard deviations
below the mean.

TABLE 12. z-SCORES FOR THE CLASS SIZE DATA

Deviation from about z-Score
Number of Students in a
the Mean ̅
𝒙𝒊 − 𝒙
Class (𝒙𝒊 ) ൬ ൰
(𝒙𝒊 − 𝒙
̅) 𝒔
46 2 2/8 = 0.25
54 10 10/8 = 1.25
42 -2 -2/8 = -0.25
46 2 2/8 = 0.25
32 -12 -12/8 = -1.50
𝑠 = 8, 𝑥ҧ = 44
 Its application in the business and accounting is that Z-score is essentially a
measure of "how weird" or "how normal" a piece of data is compared to the average. It is
a tool for standardization that allows managers to compare performance, risk, and
efficiency across different departments, time periods, or even entirely different
companies. For example, perhaps the most famous use in finance is the Altman Z-Score.
It combines five different financial ratios to predict the probability that a company will go
bankrupt within two years.
➢ Z > 2.99. "Safe" Zone (The company is healthy).
➢ 1.81 < Z < 2.99. "Grey" Zone (At risk of failure).
➢ Z < 1.81. "Distress" Zone (High probability of bankruptcy).

In manufacturing sector, Z-scores are used to measure how many defects a

process produces. A "Six Sigma" process means that the nearest "failure point" is 6
standard deviations (a Z-score of 6) away from the mean. This results in only 3.4 defects
per million opportunities. For fraud and anomaly detection, Auditors use Z-scores to spot
outliers. If an employee's expense report has a Z-score of +4.0, it is so far outside the
"normal" range of spending that it is automatically flagged for a manual audit to check for
fraud or errors.
TABLE 13. SUMMARY IN INTERPRETING BUSINESS Z-SCORES
Z-Score Business Meaning Typical Action
0.0 Exactly Average Maintain current operations.
+1.0 to +2.0 Above Average / Strong Reward performance; study for best
practices.
-1.0 to -2.0 Underperforming Monitor closely; provide coaching.
Above +3.0 "Black Swan"/ Extreme Investigate if it's sustainable or a data
Success error.
Below -3.0 Crisis / Outlier Immediate intervention; check for
systemic failure.
 Excel has a function for z-score which is
=STANDARDIZE(𝑥, 𝑚𝑒𝑎𝑛, 𝑠𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛). This means that you have to find first the
value of your average and standard deviation from your data before computing the values
of the z-scores. In the example, the z-score for the first observation is
=STANDARDIZE(B2,E2,E3) where B2 is the data for the first observation, E2 is the value
of the average and E3 is the value of the standard deviation.
CHEBYSHEV’S THEOREM

Chebyshev’s Theorem enables us to make statements about the proportion of

data values that must be within a specified number of standard deviations of the mean.
The true beauty of this theorem is its versatility: unlike the "Empirical Rule" (which only
works for bell-shaped, symmetric curves), Chebyshev’s Theorem applies to any dataset,
regardless of its shape or distribution.
1
The theorem states that for any number 𝑧 greater than 1, at least (1 − ) of the
𝑧2
data values must lie within 𝑧 standard deviations of the mean. Here are the most common
benchmarks used in business:
▪ 𝒛 = 𝟐: At least 75% of the data lies within 2 standard deviations from the mean
(µ ± 2𝜎).
▪ 𝒛 = 𝟑: At least 89% of the data lies within 3 standard deviations from the mean
(µ ± 3𝜎).
▪ 𝒛 = 𝟏𝟎: At least 99% of the data lies within 10 standard deviations from the mean
(µ ± 10𝜎).
In the real world of business, data is rarely "perfect." Returns on stocks, customer
wait times, and manufacturing errors are often skewed or lumpy. Chebyshev’s Theorem
is the "Safety Net" for analysts. For risk management, if a portfolio manager doesn't know
the exact distribution of a stock's returns, they use Chebyshev to calculate the "worst-
case scenario" for how often a loss might occur.
Moreover, in accounting, auditors use it to define the "normal range" for expenses.
Even if the data is weirdly skewed, they know that anything outside 3 standard deviations
represents at most 11% of the data, making those points worth investigating as potential
outliers.
Example 13. Imagine you are a manager at a customer service center. You don't know
the shape of your "Wait Time" distribution (it’s likely skewed because many calls are short,
but a few are very long). Your data shows that the average wait times is 5 minutes and
the standard deviation is 1 minute. However, a major client demands to know “What is
the maximum percentage of callers who have to wait more than 7 minutes or less than 3
minutes?"
Solution:
➢ The range (3 to 7 minutes) is exactly 2 standard deviations from the mean So, 𝑧 =
2.
➢ Apply Chebyshev’s formula: 1 - (1 / 22) = 1 - 0.25 = 0.75
➢ This means at least 75% of your callers wait between 3 and 7 minutes.

You can confidently tell your client: "Regardless of how messy our day gets, no
more than 25% of our customers will ever wait outside of the 3–7 minutes window."
EMPIRICAL RULE

The Empirical Rule, also known as the 68 – 95 – 99.7 Rule, is a statistical

guideline used to describe the spread of data in a Normal Distribution (the classic bell-
shaped curve). Unlike Chebyshev’s Theorem, which works for any data shape, the
Empirical Rule is specific, it only applies when the data is symmetric and bell-shaped.

The rule states that for a normal distribution, nearly all data falls within three
standard deviations of the mean.
• 68% of the data falls within 1 standard deviation from the mean (µ ± 1𝜎).
• 95% of the data falls within 2 standard deviations from the mean (µ ± 2𝜎).
• 99.7% of the data falls within 3 standard deviations from the mean (µ ± 3𝜎).

The Empirical Rule is the "gold standard" for predictability and risk assessment.
For example;
➢ Investment Risk. If stock market returns are assumed to be normal, a finance
manager knows there is only a 0.3% chance (100% - 99.7%) that a "3-sigma
event" (an extreme market crash or spike) will occur.

➢ Quality Control (Six Sigma). In manufacturing, if a process is "in control,"

99.7% of products should fall within specifications. Anything outside that 3-sigma
range is flagged as a defect.

➢ Budgeting. Managers use this to set "confidence intervals." For example, "I am
95% confident our utility costs next month will be between Php 4,000 and Php
6,000."

Example 14. Suppose Amazon analyzes its delivery data for a specific city and finds
the times are normally distributed. The average is 48 hours and the standard deviation
is 4 hours. Using the empirical rule, the logistic manager can conclude;
Using the Empirical Rule, the Logistics Manager can conclude:

▪ The 68% Range. 68% of packages will arrive between 44 and 52 hours (48 ± 4).
▪ The 95% Range. 95% of packages will arrive between 40 and 56 hours (48 ± 8).
▪ The 99.7% (Almost Certain) Range: 99.7% of packages will arrive between 36 and
60 hours (48 ± 12).

If Amazon wants to guarantee delivery to customers, they might promise "Delivery

within 60 hours." Based on the Empirical Rule, they know they will be able to keep that
promise 99.7% of the time. Conversely, if a package takes 65 hours to arrive, the manager
knows this is a significant outlier (since it's beyond 3 standard deviations), suggesting a
specific breakdown in the supply chain that needs to be investigated.
TABLE 14. COMPARISON BETWEEN CHEBYSHEV’S AND EMPIRICAL RULE
Feature Chebyshev’s Theorem Empirical Rule
Distribution Any shape (Skewed, flat,
Only Symmetric/Normal
Shape etc.)
Data within 𝟐𝝈 At least 75% Approximately 95%
Data within 3𝝈 At least 89% Approximately 99.7%
Use Case When data shape is When data is verified as
unknown "Normal"

DETECTING OUTLIERS

Sometimes a data set will have one or more observations with unusually large or
unusually small values, these extremes are “outliers”. Detecting Outliers is the
process of identifying data points that are significantly different from the majority of the
observations. These "anomalies" can represent either a breakthrough (a massively
successful product), an error (a typo in a financial statement), or a threat (fraudulent
activity).

To avoid guesswork, businesses use two main mathematical standards to flag

outliers:

A. The z-Score Method. These measures how many standard deviations a point is
from the mean. Typically, any value with a Z-score greater than +3 or less than -3 is
considered an outlier. Such data values can be reviewed for accuracy and determine
whether they belong to the set.

Example 15. We refer to the z-score of in Table 12, the z-score of -1.50 shows the
fifth class size is farthest from the mean. However, this standardized values is well
within -3 and +3 guideline for outliers. Thus, the z-scores do not indicate that the
outliers are present in the class size data

TABLE 12. z-SCORES FOR THE CLASS SIZE DATA

Deviation from about z-Score
Number of Students in a
the Mean ̅
𝒙𝒊 − 𝒙
Class (𝒙𝒊 ) ൬ ൰
(𝒙𝒊 − ̅
𝒙) 𝒔
46 2 2/8 = 0.25
54 10 10/8 = 1.25
42 -2 -2/8 = -0.25
46 2 2/8 = 0.25
32 -12 -12/8 = -1.50
𝑠 = 8, 𝑥ҧ = 44
B. The Interquartile Range (IQR) Method. This is more "robust" because it uses the
median, which isn't easily skewed by the outliers themselves.

• Calculate IQR = Q3 – Q1
• Define Fences
o Upper Fence = Q3 + 1.5(IQR)
o Lower Fence = Q1 - 1.5(IQR)
• Identify. Anything outside these fences is an outlier.

Example 16. Imagine a major Philippine retail chain like SM or Robinsons analyzing the
daily transaction volume per customer during the month of December.

▪ Data: Most customers spend between ₱500 and ₱5,000 on gifts and groceries.

▪ Outlier: An automated system flags a single transaction for ₱1,200,000 at a

department store branch in Quezon City.

▪ Detection: Using the IQR Method, the upper fence was calculated at ₱15,000. Since
₱1.2M is far beyond that, it’s an outlier.

▪ Decision: The store investigates. Is it credit card fraud? Or is it a "Balikbayan" relative

using their global savings to buy appliances for their entire extended family? In the
Philippine context, these outliers often appear during the "Ber" months due to OFW
remittances.
LESSON 4. EXPLORATORY DATA ANALYSIS
Exploratory Data Analysis is the critical first step in any data project. It is the
process of "interrogating" your data to uncover patterns, spot anomalies, and check
assumptions before you apply any complex formal modeling or machine learning. If data
science is a detective story, EDA is the part where the detective walks through the crime
scene, takes photos, and looks for fingerprints before making any accusations.

In Exploratory Data Analysis, the Five-Number Summary and the Box Plot are
inseparable tools. While the summary provides the numerical "skeleton" of the data, the
box plot provides the visual "body." Together, they offer a complete snapshot of a
dataset's center, spread, and outliers without being distorted by extreme values.

THE FIVE-NUMBER SUMMARY

This is a set of five statistics that divide your data into four equal-sized intervals
(quartiles), each containing approximately 25% of the observations.

1. Smallest Value: The smallest value in the dataset (excluding outliers).

2. First Quartile (Q1): The 25th percentile; 25% of the data falls below this.
3. Median (Q2): The 50th percentile; the middle of the data.
4. Third Quartile (Q3): The 75th percentile; 75% of the data falls below this.
5. Largest Value: The largest value in the dataset (excluding outliers).

Unlike the Mean and Standard Deviation, these five numbers are robust. If you
have one massive outlier, the Median and Quartiles won't move much, whereas the
Mean would shift significantly.

25,000 25,600 26,500 | 26,500 26,500 27,800 | 28,000 28,500 30,000 | 30,500 32,000 35,000
Q1 = 26,500 Q2 = 27,900 Q3 = 30,250

We have the data of the monthly salary of the graduates in a business school,
where we compute the Q1, Q2 , and Q3 values. Thus, the five-number summary for the
salary data is 25000, 26500, 27900, 30250, 35000. Approximately one fourth or 25% of
the observations are between these adjacent numbers in a five number summary.
THE BOX PLOT (BOX-AND-WHISKER)

A box plot is the graphical representation of the Five-Number Summary. It is the

most efficient way to see if your data is skewed and if you have outliers.

Anatomy of a Box Plot

• The Box: Drawn from Q1 to Q3. The length of this box is the Interquartile Range
(IQR), representing the middle 50% of your data.
• The Median Line: A vertical line inside the box. If it’s not in the center of the box,
your data is skewed.
• The Whiskers: Lines extending from the box to the Min and Max.
• Outliers: Individual dots plotted beyond the whiskers.

In Excel, to display the boxplot of the data of the monthly salary, fist highlight the
data, and click the “Insert” tab. Then, click the “Recommended Charts” icon, and the
“Insert Chart” will pop up. Click “All Charts” and find the “Box and Whisker” and click “OK”
button.
 By clicking the “OK” button, a box plot figure will show. But the figure needs to be
enhanced by double clicking the values in the left at the excel worksheet. A “Format Axis”
will show at the right most of the excel worksheet, search the “Axis Options” and edit the
Minimum and Maximum to the desired number values.

This is the final output showing the box plot of the monthly salary data

Max =35,000

Q3 =30,250

IQR
Q2 =27,900
Q1 =26,500

Min =25,000