TU Dortmund

Introductory Case Studies

Project 1 : Descriptive analysis of

demographic data

Lecturers:
Prof. Dr. Sonja Kuhnt
Dr. Paul Wiemann
Dr. Birte Hellwig
M. Sc. Hendrik Dohme

Author: Sharath Kumar Sadari

Group number: 2

Group members: Mridul Varghese Koshy, Gitto Benny

November 12, 2021

Contents

1. Introduction 1

2. Problem statement 1
2.1. Dataset and Data Quality . . . . . . . . . . . . . . . . . . . . . . . . . . 1
2.2. Project Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

3. Statistical methods 2
3.1. Statistical Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
3.2. Statistical Visualization . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

4. Statistical analysis 9
4.1. Frequency distribution of all variables . . . . . . . . . . . . . . . . . . . . 9
4.2. Bivariate correlation between variables . . . . . . . . . . . . . . . . . . . 10
4.3. Nature of the variables distribution (Homogeneous/Heterogeneous) . . . 11
4.4. Trend analysis between the year 2001 and 2021 . . . . . . . . . . . . . . 13

5. Summary 14

Bibliography 16

A. Appendix 17
1. Introduction

The life expectancy and fertility rate are two vital topic which every individual is con-
nected to. Because they are closely interlinked with various national interests including
economy, health care, population control, human resources, business, and employment.
Life expectancy is an empirical and statistical measure of the expected lifespan of an
organism based on different criteria such as year of birth, age, sex, demo-graphical fac-
tors, social and economical factors. The fertility rate is the measure of the number of
childbirths for every group of 1000 women of reproductive age (approximately 19 - 49).
Since, life expectancy and fertility rate are intertwined with the demography and geo-
graphical locations, visually analyzing both the variables aids policy-making especially
for the improvement of healthcare systems and economic status.
Demographic data from International Data Base (IDB U.S. Census Bureau, 2021) was
used for the analysis of life expectancy and fertility rates. The structure of the census
data used for the analysis and the objectives of the project are described in detail in
section 2.1 and section 2.2 respectively. In Section 3, we explain the statistical measure
such as mean, median, quartile, variance, standard deviation, and correlation. The
same section also explains different statistical visualization methods such as histograms,
correlation heatmaps, boxplots, and scatterplots which are used in our project to provide
visual insights into the data. In section 4, we briefly perform descriptive analysis with
the interpretation of the data to understand the frequency distribution of the variables,
bivariate correlations between variables, nature of variables distribution, and also trend
analysis between the years 2001 and 2021. Finally, the main results of the demographic
analysis of the data are summarized in section 5. (U.S. Census Bureau, October 2021)

2. Problem statement

2.1. Dataset and Data Quality

Demographic data from International Data Base (IDB U.S. Census Bureau, 2021) main-
tained by U.S Census Bureau was used for the analysis of life expectancy and fertility
rates. This sample dataset includes information about life expectancy and fertility rates
from 228 countries for the years 2001 and 2021. The dataset contains a small extract from
the IDB. The IDB is updated regularly to provide the information required for research,

1
program planning, and policymaking in the United States and globally. The sample data
extract has 456 records, 2 records each for 228 countries. The variables of the dataset
are described in the Appendix A on page 18 in Table 5. The data also has missing
values for the variables life.expectancy.females, total.fertility.rate,life.expectancy.males,
and life.expectancy.both.sexes. The missing values are from the year 2001. We will re-
place the missing values with the mean of their respective variable values. (U.S. Census
Bureau, October 2021)

2.2. Project Objectives

In this descriptive analysis project, we compare both life expectancy and fertility rates
against different countries, regions, and subregions. We use correlation matrix and
heatmaps to compare co-relation between the factors, this helps us understand the
inter-dependency between variables. Histograms are used to visualize the frequency
distribution of the dataset variables. To determine if the nature of the variable distri-
bution is homogeneous or heterogeneous between subregions we use boxplots. We also
use histograms to visualize the trend change analysis between the years 2001 and 2021.

3. Statistical methods

In this section, we present descriptive statistical methods used for our analysis of the
data. And also, we present the statistical measures and their respective formulas. We
use these methods further for analyzing the dataset. We use R (R Core Team, 2021)
running on R studio (RStudio Team, 2019) software for all the computation and visu-
alization. We use R packages bit, bit64, data.table, dplyr, zoo, tidyr, pryr, ggplot2,
ggpubr, hrbrthemes, viridis,ggcorrplot, Hmisc.

3.1. Statistical Measures

For understanding the data, we need to first know the following statistical measures and
what they represent.

2
Mean

The mean(average) is one measure of central tendency. A dataset’s mean is computed

by summing up all the values of a variable and dividing by the total observations in
the variable set. Since the mean considers all the observations, it is important to note
that it is sensitive to extreme values on only one tail of the data. Suppose we have
values x1 , x2 ....., xn in the dataset, then mean is denoted by x̄, where n denotes the total
number of observations. (Hay-Jahans, 2018)(p. 73-74)

n
1X
x̄ = xi
n i=1

Median

The median of data refers to the middle point in a dataset such that at least half of
the data points are smaller than the median and at least half of the data points are
larger. The median is always computed on a sorted dataset. If dataset X contains
values x1 , x2 ....., xn with n observations, and n is odd, then the median is the middle
point. If n is an even number, then the median will be the mean of two middle points.
(Hay-Jahans, 2018)(p. 75-76)


X n+1 , if n is odd
[ 2 ]





M ed(X) = X + X

 [ n2 ] [ n2 ]+1
, if n is even



2

Quartile

The ordered dataset values are divided into four equal sections called quartiles. A quar-
tile is a type of Quantile. The first quartile(Q1 ) covers the first 25% of the data, middle
quartile(Q2 ) covers up to 50% of the data from min point, it is also known as median,
third quartile(Q3 ) is at 75% of the data. If a dataset X contains values x1 , x2 ....., xn with
n observations arranged in ascending order, then the range R is computed by taking the
difference between the maximum and the minimum data points.

R = xmax − xmin

3
The Inter-Quartile range (IQR) is computed by taking the difference between Q3 and
Q1 .
Inter Quartile Range(IQR) = Q3 − Q1

(Hay-Jahans, 2018)(p. 77)

Variance and Standard Deviation

When we consider numeric data then there are some measures of spread. Lets consider
sample numeric data x1 , x2 ....., xn in the dataset.
Variance in a dataset is the measure of spread. It indicates how far the data is spread
out from the mean of the dataset. It considers squared distance to compute the measure
of spread. The variance S 2 is given by:

(xi − x̄)2
Pn
2 i=1
S =
n−1

Where x̄ is the mean of observations, xi is the ith observation, and n denotes the total
number of observations.
Standard Deviation (S) is another measure of spread that helps in measuring the degree
of dispersion of data points. The higher the value of S, the sparser the data is which
implies that the spread of the data is wider from the mean point and vice versa. The
square root of Variance gives us the value of Standard deviation. The Standard Deviation
is given by: v
u n (x − x̄)2
uP
t i=1 i
S=
n−1

(Hay-Jahans, 2018)(p. 76-77)

Correlation

Correlation is the measure of how two variables are related/dependent on each other lin-
early. If two variables are correlated, if one of the variables changes, then the correlated
variable also changes at a constant rate. Correlation is measured using a unit-free value
called Correlation Coefficient, the value of the correlation coefficient ranges from -1 to
1 and is denoted by r.

4
Suppose the dataset contains two variables X and Y. Let x1 , x2 ....., xn be the observa-
tions recorded for the variable X, then the mean value of X is denoted by x̄. Similarly
y1 , y2 ....., yn be the observations recorded for the variable Y, then the mean value of Y
is denoted by ȳ. The measure of the correlation between the variables X and Y is given
by the formula below:
Pn
i=1 (xi − x̄)(yi − ȳ)
r = qP
n
(xi − x̄)2 (yi − ȳ)2
Pn
i=1 i=1

Where n corresponds to the number of observations and i corresponds to ith observation.

If r tends to 0, then linear dependency is weaker. If r is positive, then the linear
dependency is positive. If the value of one variable changes, then the value of the
other variable changes simultaneously at the same rate. If r is negative, then the linear
dependency is negative. If the value of one variable increases then the value of the other
variable decreases and vice versa. (Hay-Jahans, 2018)(p. 321-322)

3.2. Statistical Visualization

The statistical measures mentioned in Section 3.1 can be visualized using descriptive
statistical methods, which puts forth an in-depth perspective of the data used. Below
are the methods used for the analysis of the dataset "census_2021_2001".

Histogram

The histograms are one of the most basic statistical methods for a quick lookup of the
data frequency distribution on grouped data. It is important to note that histogram
works only with numerical continuous data.
The construction of a histogram is straightforward. The first step is to bin/buckets the
x-axis range, that is we divide the x-axis range into continuous intervals, then count how
many data points fall into that interval. Bins should be non-overlapping and continuous
on the interval. Setting the right number of bins can provide more details about the
frequency distribution. The width of the bins need not be the same. When a probability
density curve is placed on a density histogram, it is feasible to determine (roughly) what
type of probability distribution the underlying random variable has. Figure 3.2.1 a
density based frequency distribution histogram.

5
 Figure 3.2.1: A density based frequency distribution histogram, for reference.

The y-axis lists the frequency value. A rectangular bar is plotted with a height equivalent
to the value of the frequency. Histogram can be used to depict relative frequencies and
density estimation. If x-axis interval lengths are all 1, then the histogram is a relative
frequency graph. (Hay-Jahans, 2018)(p. 131-137)

Correlation heatmap

A heatmap is a data visualization technique that uses variation in color from a color scale
to show the correlation between the variables in a visually appealing way. By observing
the color change on the heatmap, we can visualize easily whether the variables are
correlated and to what degree are the variables correlated. We can also observe patterns
in values of the variables plotted on the respective axis. Figure 3.2.2 is an example for
heatmap (Hoffman, 2020).
The graph is constructed from the idea of correlation which is described in section 3.1
under statistical measures. The variables are plotted on each X and Y-axis. The size
of the cell is arbitrary but large enough to be visible. Many different color schemes can
be used to illustrate the heatmap. The darker the color stronger is the correlation. In
Figure 3.2.2, the dark red color indicates a stronger negative correlation with r value -1
and the color fades as it reaches 0, indicating the negative correlation is getting weaker.
On the contrary, the dark blue color indicates a stronger positive correlation with r value
+1 and the color fades as it reaches 0, indicating that the positive correlation is getting
weaker. On a scale of 0, which is colorless, there is no correlation between the variables.
(Hay-Jahans, 2018)(p. 321-322)

6
 Figure 3.2.2: A Correlation heatmap, for reference.

Boxplot

Boxplot is a standardized statistical method that provides a visual five-number sum-

mary of numerical data (grouped) using ”Quartiles”, which helps us understand data
distribution. It is built of two parts, a box, and whiskers. The five-point summary is the
minimum value, First Quartile (Q1 ), Middle Quartile (Q2 or Median), Third Quartile
(Q3 ) and maximum value.

Figure 3.2.3: A boxplot, for reference.

A box is constructed between Q1 and Q3, with a horizontal line in the middle showing
median. Whisker is the line from the box to the minimum and the maximum value.

7
Any observations that lie beyond 1.5*IQR are referred to as outliers. Figure 3.2.3 is an
example for boxplot. The boxplots help in understanding the data distribution. Based
on the length of the whisker we can understand the skewness of the data. The data is
right-skewed if the right whisker is longer and left-skewed if the left whisker is longer.
If the median is at the center of the box and both the whisker of equal length, then the
distribution is symmetric. (Hay-Jahans, 2018)(p. 137-146)

Scatterplot

A scatterplot (also called a scatter graph, scatter chart, scattergram, or scatter diagram)
is a 2-dimensional plot, which typically plots two variables (dependent variable Y, and
independent variable X) as pair (Xi ,Yi ) where i=1, 2,..., n using Cartesian coordinates.
Each pair represent the position of observation along the horizontal and vertical axis.
Figure 3.2.4 is an example for scatterplot.

Figure 3.2.4: A scatterplot, for reference.

From the scatterplot, we can find the correlation between the variables with a certain
confidence. Based on the slope, the association can be positive association, uncorrelated
or negative association. If the slope of the plotted points tends to increase, then the
variable is positively correlated. For a decreasing slope, the correlation is negative. We
can include a third variable by including color/shape for points plotted. (Hay-Jahans,
2018)(p. 159-169)

8
4. Statistical analysis

The statistical methods outlined in section 3 are applied to the dataset in this section,
and the findings are interpreted.

4.1. Frequency distribution of all variables

We start by looking at the frequency distributions of the variables "Life expectancy both
sexes", "Total Fertility rate", "Life expectancy, and "Life expectancy male".

Figure 4.1.1: A Density based histogram, for frequency distribution of variables.

The graph is plotted for the data for the year 2021. From the Figure 4.1.1, "Total fertility
rate", we observe a higher concentration of data between 1 and 3, with the maximum
value recorded at around 1.8. The distribution is left-skewed, with a mean value of
around 2.5. The frequency distribution of Life expectancy is right-skewed.
From the Figure 4.1.1, the mean value of "Life expectancy of both Genders" lies at
around 74.5, which is more than the mean value of "Life expectancy of male" which is
around 72, and less than the mean of "Life expectancy of female" which is close to 77.
The Figure 4.1.2 overlays both male and female life expectancy on a single plot for a
better understanding of differences between the sexes. From this graph, it is evident
that females, in general, have a higher life expectancy than men.

9
Figure 4.1.2: A Density based histogram, for frequency distribution Male Vs Female Life
Expectancy.

For a better understanding of the variations between the sexes, in Figure 4.1.2 we overlap
both male and female life expectancy on a single graph. We observe, that Females, on
average, have a longer life expectancy than men, as shown in this graph.

4.2. Bivariate correlation between variables

Using a heatmap, we check for bivariate correlation between the variables. The cor-
relation for the Total Fertility rate, Life expectancy of Males and Females, and Life
expectancy of both sexes is computed. The correlation value on the color scale ranges
from [-1,1]. We use a blue-red color scheme. The positive correlation (r>0) is color
coded using shades of blue, and the negative correlation (r<0) using shades of red.
Figure 4.2.1 plotted for the data for the year 2021. It is evident that the Fertility
rate has a negative correlation value of -0.79, -0.77, and -0.81 with the variables Life
expectancy of both sexes, the Life expectancy of Males and Females respectively, and
vice-versa. The variables Life expectancy of both sexes has a positive correlation with
the Life expectancy of Males and Life expectancy of Females with correlation value 0.99
and 0.99 respectively. The Life expectancy of Males has a positive correlation with the
Life expectancy of both sexes and the Life expectancy of Females with correlation values
0.99 and 0.97 respectively. The Life expectancy of Females has a positive correlation
with the Life expectancy of both sexes and the Life expectancy of Males with correlation
values 0.99 and 0.97 respectively.

10
 Figure 4.2.1: A Heatmap among variables.

4.3. Nature of the variables distribution

(Homogeneous/Heterogeneous)

The boxplot in the Figure 4.3.1 titled "Life expec of females w.r.t to Sub Region" and
"Life expec of males w.r.t to Sub Region" shows that Africa is completely heterogeneous
compared to their peer regions. The life expectancy of females and males in Africa is
smaller than their counterparts in other regions.
These plots reveal that the distribution of data is heterogeneous between subregions
and homogeneous within subregions. In the American region Northern America, in
the European region, Northern Europe, and Western Europe, and the Oceania region
Australia/New Zealand are eminently homogeneous subregions. The Australia/New
Zealand subregion is the most homogeneous, with no single extreme value and a very
small whisker line, implying that life expectancy for females and males in the sample
population is almost identical with a median value of 84 for females and 81 for males. In
Asia, all the subregions are heterogeneous. When we compare the median value in each
subregion for males and females we observe that the median value for females is higher
than that of males. In Eastern Africa, the median value for females is around 68 and for
males, it is around 64. In Central America, the median value for females is around 79
and for males, it is around 73. With this, we can conclude that female life expectancy
is higher than male life expectancy in all subregions.

11
 Figure 4.3.1: The Boxplots to visualize the distribution of variables.

Further, when we analyze the graph from Figure 4.3.1 with the title "Life expec of both
sexes w.r.t to Sub Region", it is similar to the graphs for the values for females and
males only and has similar data distribution.
In addition, we compare the total fertility rate of the subregions in the Figure 4.3.1 titled
as "Total fertility rate w.r.t to Sub Region". The African subregions have a substantially
higher fertility rate. However, they vary within African subregions and are less ho-
mogeneous within each subregion. Within their subregions, Southern Europe, Western
Europe, Australia/New Zealand, and Eastern Europe are all homogeneous. The Ameri-
cas, Oceania, and Europe subregions are less heterogeneous between subregions because
their variability is lower. Western Africa has the highest fertility rate of around 7 in any
subregions. In comparison to other subregions, the Asian subregion has more extreme

12
values, with a median fertility rate of about 1.2, Eastern Asia has the lowest fertility
rate. Since it was previously observed that the life expectancy is inversely proportional
to the fertility rate, it can be observed that Africa has a higher fertility rate and lower
life expectancy compared to the other regions. The data distribution in other regions is
nearly homogeneous, with few exceptions within the subregions. Further, the boxplot
summaries are provided in the Appendix A on page 17 in Table 1-4.

4.4. Trend analysis between the year 2001 and 2021

Data for the year 2021 was further compared to the data from 2001 using histograms
and scatterplots. Figure 4.4.1 shows the changes in the parameter values over the years.
The skewness of the data is almost similar. Over the last 20 years, the average life
expectancy has increased by about 6 years. In contrast, the mean fertility rate reduced
from 3 in the year 2001 to approximately 2.5 in the year 2021.

Figure 4.4.1: A Histogram, for trend analysis for the year 2001 and 2021.

From the Figure 4.4.2, we observe that in comparison to the trend we saw earlier for
the year 2021 vs 2001, there are several exceptions in both life expectancy and fertility
rate. In the last 20 years, in the African region, for South Sudan and Sudan we observe
a decrease in life expectancy. In contrast, we observe an increase in life expectancy by
approximately 20 years in countries like Zimbabwe, Sierra Leone, Zambia, and Uganda,
compared to a global increase of 5 years. Bucking the trend, we observe an increased

13
fertility rate in Sudan and South Sudan because of the negative correlation between life
expectancy and fertility rate. In contrast to the gradual decline in total fertility, Asian
countries such as Afghanistan, Timor-Leste, and Yemen have seen their fertility rates
drop drastically.

(a) Life Exp Male (b) Life Exp Female

(c) Life Exp Both (d) Fertility Rate

Figure 4.4.2: The scatterplots of different variables of year 2001 vs 2021 color coded by
region

5. Summary

In this project, we used a sample dataset by the International Data Base(IDB), main-
tained by the U.S Census Bureau, for Fertility and Life expectancy. We wanted to
understand, the variables distribution, the variables dependency, and the trend of the
Fertility and Life expectancy of 228 countries for the years 2001 and 2021.
Using Histogram, we analyzed the distribution of the variable in the dataset. "Life
expectancy of both gender", "Life expectancy of Male", "Life expectancy of female" are
all left skewed distributions, whereas "Total fertility rate" is a right skewed distribution.
The empirical distribution is concentrated between 1 and 3 for the total fertility rate.

14
The mean life expectancy of females is higher than males which are approximately 77
and 72 respectively.
The variables correlation was depicted using the correlation heatmap, the variables under
consideration for correlation were "Life expectancy of both gender", "Life expectancy of
Male", "Life expectancy of female", and "Total fertility rate". The "Total fertility rate"
has a negative correlation with Life expectancy. And all the life expectancy variables
are mutually positively correlated.
To understand, the variable distribution, we used the boxplot and the scatterplot to
visualize the distribution. We found that the Africa region is heterogeneous to other
regions, in terms of both Life expectancy and Fertility rate. Within the regions, some
subregions are heterogeneous. Using the scatterplot, we could also find a few of the
outliers for the Africa, Asia, and Europe regions.
To understand the trend between the years 2001 and 2021, we compared the value of
each variable in the year 2021 to its value in the year 2001. The scatterplots were used to
visualize this trend analysis. All the variables show a linear relationship with a positive
slope. Since life expectancy and fertility rate have a negative correlation, we can observe
this effect on the increased life expectancy for both females and males and decreased
fertility rate over 20 years.
To conclude, economically prosperous countries such as European countries, North
America, in the Asia continent Japan and Oceanic region Australia and New-Zealand
have lower fertility rate and higher life expectancy than African countries, Asia (except
Japan), South American countries. Both high life expectancy and high fertility rates are
good. However, the scenario of high life expectancy with a low fertility rate can poten-
tially increase the average age of the population, and the share of the elderly population
increases. For example, the case of Japan, which has a high life expectancy and a low
fertility rate has an aging population which is resulting in a shortage of human resources.
Which potentially can harm the economy, industry output decreases, inflation increases,
and also puts the health sector under immense stress.
For future studies, it might be interesting to work with the data by including several
attributes like education, salary, nutrition level, health index, age, and more. It would
bring in much more detailed insight into how these attributes are interlinked. How
does the nutrition level impact life expectancy and fertility rate, or does education level
help in a healthy lifestyle? and would a healthy lifestyle in combination with high life
expectancy reduce the number of deaths due to covid? and so on.

15
Bibliography
Christopher Hay-Jahans. R Companion to Elementary Applied Statistics. CRC Press,
Taylor Francis Group, Boca Raton., 2018. ISBN 978-1-4614-7137-0.

Kat Hoffman. Customizable correlation heatmaps in r using purrr and ggplot2.

https://towardsdatascience.com/customizable-correlation-plots-in-r-b1d2856a4b05,
2020. [Online; accessed 11-November-2021].

R Core Team. R: A Language and Environment for Statistical Comput-

ing. R Foundation for Statistical Computing, Vienna, Austria, 2021. URL
https://www.R-project.org/.

RStudio Team. RStudio: Integrated Development Environment for R. RStudio, Inc.,

Boston, MA, 2019. URL http://www.rstudio.com/.

U.S. Census Bureau. International Data Base: Population Estimates and

Projections Methodology. U.S. Census Bureau, October 2021. URL
https://www.census.gov/programs-surveys/international-programs/about/idb.html.
[Online; accessed 11-November-2021].

16
A. Appendix

A Additional tables

Table 1: Table Summary for Life Expectancy Male for all region

Region Min Q1 Median Mean Q3 Max

Africa 37.35 53.03 59.64 59.53 65.47 77.34
Americas 55.95 69.90 72.88 72.44 75.39 81.32
Asia 44.85 66.69 70.70 69.93 74.24 83.48
Europe 59.64 71.72 75.78 74.73 78.79 85.55
Oceania 57.50 66.72 71.61 70.48 74.18 80.73

Table 2: Table Summary for Life Expectancy Female for all region

Region Min Q1 Median Mean Q3 Max

Africa 42.14 56.10 63.37 63.23 69.22 83.31
Americas 57.69 74.36 78.20 77.50 81.05 86.03
Asia 46.83 71.41 75.60 74.74 78.98 89.05
Europe 70.21 78.62 81.50 81.17 83.97 93.40
Oceania 62.97 71.26 76.23 75.23 79.19 85.17

Table 3: Table Summary for Total Fertility rate for all region

Region Min Q1 Median Mean Q3 Max

Africa 1.531 3.361 4.517 4.397 5.454 8.110
Americas 1.200 1.795 2.019 2.197 2.470 4.404
Asia 0.8561 1.7512 2.2409 2.5669 2.9552 7.5099
Europe 1.083 1.368 1.521 1.579 1.743 2.807
Oceania 1.700 2.174 2.670 2.845 3.425 4.621

Table 4: Table Summary for Total life expectancy both sexes for all region

Region Min Q1 Median Mean Q3 Max

Africa 39.71 54.61 61.56 61.35 67.19 80.25
Americas 56.81 72.24 75.40 74.92 78.08 83.62
Asia 45.81 68.89 72.45 72.27 76.42 86.19
Europe 66.06 75.25 78.30 77.85 81.45 89.40
Oceania 60.70 68.64 73.91 72.79 76.42 82.89

17
 Table 5: Summarizes the variables of dataset

Variable Name Description Data Type

country Country Name Categorical
genc Country Code Categorical
subregion smaller part of continent Categorical
region Continent Categorical
year Year in which the data is recorded Categorical
total.fertility.rate The number of children a women Continuous
would have in her entire life.
life.expectancy.both.sexes A newborn’s life expectancy, as- Continuous
suming that population death rates
remain constant.
life.expectancy.males A male newborn’s life expectancy, Continuous
assuming that population death
rates remain constant.
life.expectancy.females A female newborn’s life expectancy, Continuous
assuming that population death
rates remain constant.

18