Vol.

9, 2024-11

Randomistic Data Elements

Hugo Hernandez
ForsChem Research, 050030 Medellin, Colombia
[email protected]

doi: 10.13140/RG.2.2.22772.36485

Abstract
Randomistics refers to the integration of the deterministic and random realms into a single
world. In this report, the general concept of randomistics will be discussed, considering all
types of data elements. On one hand, it applies to either changing or unchanging data
elements, which will be denoted as Variables and Invariants, respectively. Randomistics also
applies to any type of data element, according to the nature of the values contained. In this
sense, numerical/quantitative (either discrete or continuous) or categorical/qualitative
randomistic data elements are discussed in detail, highlighting their main differences.
Particularly, numerical randomistic data elements are characterized by special operators
involving mathematical operations of the data element values, including the expected value
operator, moment operators, the variance operator, and many others. Only a limited set of
functions applies to categorical data elements. However, when the outcome of those functions
is numerical, all mathematical operators can now be employed.

Keywords
Categorical, Complex, Continuous, Discrete, Events, Functions, Imaginary, Invariants,
Numerical, Probability Density, Probability Distribution, Randomistic, Standard, Variables

1. Introduction

The realm of randomistic variables can be considered as the combination of both random
(uncertain) and deterministic (certain) variables [1,2]. In some way, randomistic variables are
analogous to complex variables comprising both real and imaginary terms. Similarly,
randomistic variables can be considered as combinations of both deterministic and random
terms [2].
Most of the previous reports about randomistic variables have focused on discussing
continuous, real, numerical variables. However, other types of variables, particularly discrete

Cite as: Hernandez, H. (2024). Randomistic Data Elements. ForsChem Research Reports, 9, 2024-11, 1-22.
doi: 10.13140/RG.2.2.22772.36485. Publication Date: 24/09/2024.
 Randomistic Data Elements
Hugo Hernandez
ForsChem Research
[email protected]

variables (either numerical or categorical), and non-real variables (imaginary and complex) can
be considered randomistic variables. In addition, invariants, which are the opposite of variables
and represent another type of data elements, can also be considered randomistic.
Thus, the purpose of this report is introducing a more general framework of randomistic data
elements, valid for any arbitrary type of data element. Section 2 introduces some basic
definitions for understanding the difference between deterministic, random and randomistic
data elements. Section 3 discusses quantitative or numerical randomistic data elements,
including discrete and continuous data elements, as well as real, imaginary and complex data
elements. Section 4 describes qualitative or categorical randomistic data elements. Finally,
Section 5 discusses in more detail the concept of invariance of data elements.

2. Basic Definitions
In this Section, the basic definitions of deterministic, random and randomistic data elements
are presented. Additional basic definitions are included for a better understanding of these
concepts.

Data Element: A data element is a piece of information about a population (comprising a set of
individuals) obtained under certain observation conditions. The information can
be either quantitative or qualitative. Data elements can be invariant or variable.
The data element is commonly represented by a mathematical symbol (a character
in any script, a digit, or a string of characters and/or digits).
Quantitative Data Element: Data element where the information contained corresponds to
mathematical numbers.
Qualitative Data Element: Data element where the information contained does NOT correspond
to mathematical numbers. The information may contain non-mathematical
sequences of digits, or sequences of characters and/or digits, or it may contain
images or any other form of representation of the specific properties of a set of
elements.
Mathematical Number1: Sequence of digits which can be subjected to basic mathematical
operations with coherent results.
Invariant Data Element: Data element whose value (quantitative or qualitative) never changes for
the particular system considered, even when the observation conditions may
change. Invariant data elements can be constants or parameters. It will be simply
denoted as invariant.

1
Notice that sequences of digits can also be used as categories. In that case, basic mathematical
operations with those numbers (such as addition, product, etc.) do not yield coherent results.

24/09/2024 ForsChem Research Reports Vol. 9, 2024-11

10.13140/RG.2.2.22772.36485 www.forschem.org / t.me/forschem (2 / 22)
 Randomistic Data Elements
Hugo Hernandez
ForsChem Research
[email protected]

Constant: Invariant data element whose value (quantitative or qualitative) is known (or
assumed to be known), with absolute certainty.
Parameter: Invariant data element whose value (quantitative or qualitative) presents any
degree of uncertainty or is completely unknown.
Variable Data Element: Data element whose value (quantitative or qualitative) may change in the
particular system considered when the observation conditions change. It will be
simply denoted as variable.
Deterministic Data Element: Data element whose value (quantitative or qualitative) is known (or
assumed to be known), with absolute certainty. Thus, according to this definition,
a constant is a deterministic invariant. Variables can also be deterministic if the
different values obtained for all observation conditions are known with certainty.
Deterministic data elements will be represented by italic letters (e.g. ).
Random Data Element: Data element whose value (quantitative or qualitative) presents any
degree of uncertainty or is completely unknown. Thus, according to this definition,
a parameter is a random invariant. Variables can also be random if the value
obtained for at least one observation conditions is uncertain. Random data
elements will be represented by handwritten letters (e.g. ).
Randomistic Data Element: Data element whose value (quantitative or qualitative) can be known
with absolute certainty, known with partial certainty, or completely unknown.
According to this definition, any data element can be considered a randomistic
data element. Randomistic data elements will be represented by double line
letters (e.g. ).

The purpose of introducing the concept of randomistic data element here is simply to indicate
that both deterministic and random realms can be integrated into a single realm governed by
the exact same rules. Particularly, a non-exhaustive list of those common rules include the
following:

 Randomistic data elements can be associated to properties of real objects, or to

abstract entities, subject to specific conditions or constraints. The real objects or
abstract entities, in turn, represent sets containing either a single or multiple elements2
or individuals3 satisfying those conditions or constraints. These sets are commonly
denoted as populations.

2
An element of a population should not to be confused with a data element.
3
An individual in a population is not restricted to a biological or physical entity. The population may
represent, for example, a set of different observations of the same property of the same object. In this
case, each observation will be considered as a different individual of the population.

24/09/2024 ForsChem Research Reports Vol. 9, 2024-11

10.13140/RG.2.2.22772.36485 www.forschem.org / t.me/forschem (3 / 22)
 Randomistic Data Elements
Hugo Hernandez
ForsChem Research
[email protected]

 Randomistic data elements can be evaluated for each element or individual of the
population, resulting in a realization of the data element. Since a realization represents
a known value of the data element, each realization can be considered as a constant.
 Randomistic data elements can be transformed by means of functions. The input
arguments of the function can be numerical or categorical. Similarly, the output of the
function can also be numerical or categorical, regardless of the nature of the input
arguments.
 Randomistic data elements can be characterized by a probability distribution function.

Function: A function is a set of rules for the assignment of values to a numerical or

categorical outcome data element, depending on the values of one or more data
elements used as input arguments. The set of rules is typically described by logical
and mathematical operations. In general we have:
( )
(2.1)
While mathematical operations are exclusive of numerical data elements, logical
operations can be applied to both numerical and categorical data elements.
Probability Distribution Function: A function that uses a realization value of a randomistic data
element as input argument, and yields as outcome the probability of occurrence
of such realization value in the population associated to such data element. The
probability of occurrence is simply the fraction of elements or individuals in the
population having the same realization value considered. The probability
distribution function of realization in the randomistic data element will be
represented by ( ) .

Unfortunately, it is usually unfeasible to know the values of all elements in a population, so the
probability distribution function will be unknown. For that reason, probability distribution
models are commonly employed to represent the probability distribution function of a
randomistic data element. A model is an approximate representation of reality. However, we
must be very careful to avoid confusing models with reality [3].
Also, a clear definition of the population is required to adequately identify the data element.
Ambiguity in the definition of the population typically leads to difficulties in the interpretation
of the information contained in the data elements.
Since the probability of occurrence represents a fraction, we may conclude that, as any other
fraction, it is limited to the range:

( ) ( )
(2.2)
where represents any possible realization of , and ( ) is the sample space of .

24/09/2024 ForsChem Research Reports Vol. 9, 2024-11

10.13140/RG.2.2.22772.36485 www.forschem.org / t.me/forschem (4 / 22)
 Randomistic Data Elements
Hugo Hernandez
ForsChem Research
[email protected]

Sample Space: The sample space ( ( )) of a randomistic data element is the set of all possible
realization values ( ) of .

Notice that the sample space is different from the population of individuals (or elements).
Furthermore, the size of the sample space ( ( ) ) is typically different from the size of the
population ( ), both defined as follows:

Population Size: The population size ( ( ) ) of a randomistic data element is the total number
of individuals (or elements) belonging to the data element.

Sample Space Size: The sample space size ( ( ) ) of a randomistic data element is the total
number of different realization values observed in the population of the data
element.

The probability distribution function can then be described as follows:

( )
( )
( )
(2.3)
where ( ) is the number of individuals in the population of having a realization value .

Notice that:
( )

∑ ( ) ( )

(2.4)
From Eq. (2.3) and (2.4), an important property of the probability distribution function (or a
probability distribution model) of any randomistic data element can be derived:
( )

∑ ( )

(2.5)
Now, we can use the probability distribution function to distinguish deterministic from random
data elements, since the maximum value of the probability distribution function of a
deterministic data element ( ) is exactly :

( )
( )

(2.6)
whereas the maximum value of the of the probability distribution function of a random data
element ( ) is less than :

24/09/2024 ForsChem Research Reports Vol. 9, 2024-11

10.13140/RG.2.2.22772.36485 www.forschem.org / t.me/forschem (5 / 22)
 Randomistic Data Elements
Hugo Hernandez
ForsChem Research
[email protected]

( )
( )

(2.7)
Even when some common rules apply to all randomistic data elements, specific properties are
observed depending on the nature of the observable realization values. According to the
nature of realization values, let us classify randomistic data elements into: Numerical or
Categorical. Numerical data elements can also be classified either into discrete or continuous,
or into real, imaginary or complex. The common properties of these types of data elements will
be discussed in the following sections.

3. Numerical Randomistic Data Elements

Numerical data elements are characterized because all their possible realization values are
numbers with a mathematical interpretation. That is, numbers which are subject to coherent
mathematical operations. For example, the number of leaves in a plant is a numerical data
element with mathematical interpretation. If we add the number of leaves in a group of plants,
we obtain a coherent result, as it represents the total number of leaves in the group. On the
other hand, the social security number is not a numerical data element. In this case, we can also
add the social security numbers in a group of people, but we will not obtain the social security
number of the group, but just a meaningless, incoherent result.
Thus, numerical data elements can be transformed using mathematical operations as well as
mathematical functions. Some particular functions are of special interest for numerical
randomistic data elements including:

Cumulative Probability Function: Represents the probability of occurrence of any realization value
less than or equal to the value given in the input argument, associated to
randomistic data element . It is denoted by .
( )

( ) ( ) ∑ ( ) ( )

(3.1)
where is a function defined as:

( ) {

(3.2)
Notice that necessarily requires a sense of order of the possible realization values in the data
element. For this reason, it cannot be applied to data elements whose values have no
mathematical sense of order (i.e. categorical data elements).

24/09/2024 ForsChem Research Reports Vol. 9, 2024-11

10.13140/RG.2.2.22772.36485 www.forschem.org / t.me/forschem (6 / 22)
 Randomistic Data Elements
Hugo Hernandez
ForsChem Research
[email protected]

Expected Value Operator: Represents the average value of all elements in the population. It is
denoted by ( ) .

In mathematical terms, we have:

( )

( ) ∑ ()
( )

(3.3)
where () is the realization value of the -th individual in the population of . Notice that this
operator is meaningless for categorical variables, where addition makes no sense.
Alternatively, the expected value operator can be expressed in terms of the probability
distribution function as follows:
( ) ( )

( ) ∑ ( ) ∑ ( )
( )

(3.4)
The expected value is a deterministic data element, because the individual values in the
population are assumed known.

The expected value operator can also be applied to any arbitrary function of ( ( )) as
follows:
( ) ( )

( ( )) ∑ ( ( )) ∑ ( ) ( )
( )

(3.5)

Notice that the cumulative probability function defined by Eq. (3.1) can be also interpreted as
the expected value of function ( ) (Eq. 3.2).

Moment Operator: Represents the expected value of the a-th power function of a randomistic
data element corrected by an amount b. It is denoted by ( ).

( ) ( )

( ) (( ) ) ∑( () ) ∑( ) ( )
( )

(3.6)

The data elements and are invariants. is a constant, representing the order of the
moment. While a non-negative integer is commonly employed, any real value can be used [1]. ,
on the other hand, is a pivot or correction element, and can be a constant or a parameter.

24/09/2024 ForsChem Research Reports Vol. 9, 2024-11

10.13140/RG.2.2.22772.36485 www.forschem.org / t.me/forschem (7 / 22)
 Randomistic Data Elements
Hugo Hernandez
ForsChem Research
[email protected]

Typical values of include the constant zero ( ) value (raw moments), and the expected value
parameter ( ) (central moments). Raw moments are simply denoted by ( ) .

Moment Variation Operator: Represents the difference between the a-th raw moment of a
randomistic data element and the a-th power of its first raw moment. It is
denoted by ( ) .
( ) ( )

( ) ( ) ( ) ( ) ( ) ∑ () ( ∑ ( ))
( ) ( )

( ) ( )

∑ ( ) (∑ ( ))

(3.7)

Numerical deterministic data elements ( ) are characterized because their moment variation
operator for any arbitrary order is exactly zero:

( )
(3.8)

Numerical random data elements ( ) are characterized because their moment variation
operator is exactly zero only for :

( )
(3.9)

Variance Operator: The variance operator is a particular moment operator corresponding to the
the second central moment. It is also equivalent to the second moment variation
operator. It is denoted by ( ).

( ) ( ( )) ( ) (( ) ( ) ( )
( ))

(3.10)

From Eq. (3.8) and (3.9) we can conclude that deterministic data elements have zero variance,
while random data elements have non-zero variance. Of course, for randomistic data elements
the variance may take any value.

Standard Deviation Operator: The standard deviation operator represents the positive square
root of the variance operator. It is denoted by ( ) .

( ) √ ( ) | ( ) |
(3.11)

24/09/2024 ForsChem Research Reports Vol. 9, 2024-11

10.13140/RG.2.2.22772.36485 www.forschem.org / t.me/forschem (8 / 22)
 Randomistic Data Elements
Hugo Hernandez
ForsChem Research
[email protected]

Deterministic and random data elements can also be distinguished by their standard deviation
value. If ( ) , then the data element is deterministic; otherwise it is random.

Due to the mathematical interpretation of numerical data elements, numerical randomistic

data elements ( ) can be represented as the linear combination of a deterministic data
element ( ) and a random ( ) data element (both having the same population size), as
follows:

(3.12)
where and are real invariant data elements, representing the contribution of the
deterministic and random terms, respectively.
Notice that the expected value of such randomistic variable is4:
( )

( ) ( ) ∑( () ( )) ( ) ( )
( )

(3.13)
and its second raw moment is:

( ) (( ) ) ( ) ( ) ( ) ( ) ( )

(3.14)
Thus, the variance of the randomistic data element is:

( ) ( ) ( ) (( ) ) ( ( ) ( ))

( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )

( ( ) ( )) ( ( ) ( )) ( ) ( )

(3.15)
Now, since ( ) , because is a deterministic data element, then:

( ) ( )
(3.16)
where ( ) .

This means that the randomistic data element is deterministic when (that is, when
), and is random when (that is, when ).

( ) ( )
4
Since ( ) ( ) ( ) then, we obtain: ( ) ∑ ( () ( )) ∑ ()
( ) ( )
( ) ( ) ( )
∑ () ∑ () ∑ () ( ) ( ).
( ) ( ) ( )

24/09/2024 ForsChem Research Reports Vol. 9, 2024-11

10.13140/RG.2.2.22772.36485 www.forschem.org / t.me/forschem (9 / 22)
 Randomistic Data Elements
Hugo Hernandez
ForsChem Research
[email protected]

An important conclusion is that, regardless of the value of the coefficient , the data element
is random as long as .
The randomistic data element given in Eq. (3.12) can be equivalently expressed in terms of
standard data elements as follows:

(3.17)
where and are invariant terms, and and are standard data elements. represents a
standard deterministic data element, whereas represents a standard random data element.
Notice that and when and .

3.1. Standard Data Elements

Standard data elements can be defined to facilitate the notation and operation with
randomistic data elements in general. Thus, we have the standard deterministic data element,
and standard random data elements.

Standard Deterministic Data Element: Denoted by the Greek letter (upper case ypsilon). This
data element is characterized by the following properties: ( ) , and
( ) . It behaves exactly as the constant number [2]. It can also be
described by the following probability distribution function:

( ) ( ) {
(3.18)
where ( ) represents Kronecker’s delta function [4].

The standard deterministic data element is obtained by transforming any deterministic term
( ) as follows:

( )
(3.19)
On the other hand, standard random data elements are obtained after transforming any
randomistic data element as follows5:

(3.20)

5
Since , the randomistic data element cannot be deterministic and must necessarily be a
random data element. However, it will be treated here as randomistic, as it contains both deterministic
( ) and random ( ) terms. When , only the standard deterministic data element is used.

24/09/2024 ForsChem Research Reports Vol. 9, 2024-11

10.13140/RG.2.2.22772.36485 www.forschem.org / t.me/forschem (10 / 22)
 Randomistic Data Elements
Hugo Hernandez
ForsChem Research
[email protected]

Depending on how these terms are defined, we obtain different types of standard random data
elements, as indicated in Table 1.
Table 1. Types and Properties of Standard Random Data Elements [5]
Invariants
Type Bounds of Bounds of Properties of

( ) ( )
( )
[ ) [ )
( )

I ( ) ( ) ( ) ( ) ( )
( ] ( ]
( )

( ) ( )
[ ] [ ]
( ) ( )

[ ) [ )
( )
[ ] [ ]
( )
II ( ) ( )
( ] [ )
( )
[ ] [ ]
( )

III [ ] [ ] ( ) ( )

where ( ) and ( ) represent the minimum and maximum values of the

randomistic variable .

Standard Random Data Element: Denoted by the Greek letter (upper case xi). This data element
is characterized by the following properties (depending on the type):
 Type I: ( ) , ( ) .
 Type II: ( ) , ( ) .
 Type III: ( ) , ( ) .

Other types of standard random data elements can be defined, but these are probably the
most common types found in the literature.
A single randomistic data element can be represented in terms of different standard data
elements, particularly when the sample space is bounded. For example, a randomistic variable
bounded in the finite range [ ] can be represented in terms of different standard data
elements as follows:

( ) ( ) ( ( ) ) ( ( ) )
( )
(3.21)
where . However, all these standard elements are inter-related:

24/09/2024 ForsChem Research Reports Vol. 9, 2024-11

10.13140/RG.2.2.22772.36485 www.forschem.org / t.me/forschem (11 / 22)
 Randomistic Data Elements
Hugo Hernandez
ForsChem Research
[email protected]

( ) ( )
( )( ) ( )( )
( ) ( )
( )
( ) ( )
( ) ( )
(3.22)
( ) ( )
( ) ( ) ( )
( ) ( ) ( )

( )
( )
(3.23)
( ) ( )
( ) ( ) ( )
( ) ( ) ( )

( )( )
( )
(3.24)
( ) ( ) ( )
( ) ( ) ( )

( )
( )

(3.25)
Each standard data element has its own probability distribution function. In the case of the
standard deterministic data element, it was shown that it is the Kronecker delta function. For
standard random data elements, any probability distribution function is possible as long as it
satisfies the properties defined in Table 1, as well as the inherent properties of the probability
distribution function (Eq. 2.2 and 2.5).
The probability distribution functions of standard random data elements are related to the
probability distribution function of the original randomistic data element by means of the
change of variable theorem [6], which will be discussed in the next section.

3.2. Change of Variable Theorem

The change of variable theorem simply states that the probability of a certain outcome value of
a function is the joint probability of all input argument combinations yielding such value.
Let us consider the following the outcome of a general function of randomistic data
elements ( ) used as input arguments. The outcome is also assumed randomistic.

( )

(3.26)

24/09/2024 ForsChem Research Reports Vol. 9, 2024-11

10.13140/RG.2.2.22772.36485 www.forschem.org / t.me/forschem (12 / 22)
 Randomistic Data Elements
Hugo Hernandez
ForsChem Research
[email protected]

The change of variable theorem then implies that:

( ) ∑ ( )

(3.27)
where is the number of combinations of the input arguments (solutions) yielding the same
outcome value , and is the number of different outcome values (including ) that are
obtained for each specific combination of input arguments.
Eq. (3.27) is greatly simplified when there is a single solution to the function , and when there
is a single input argument, or at least, when the input arguments are independent from each
other.

Independence: Two or more data elements are independent when the joint probability of
occurrence of any set of realization values is the product of the individual
probabilities of each realization value:

( )
∏ ( )

(3.28)

Eq. (3.28) indicates that a data element is independent from other data elements when the
probability of occurrence of any realization of the data element is not influenced by the
realization values of the other data elements.
If two or more data elements used as input arguments in a function are dependent, then they
can ultimately be expressed as functions of one or more independent data elements.
The change of variable theorem applied to linear transformations of a single variable yields a
relatively simple result. As an example, let us consider the linear transformation function
presented in Eq. (3.20). Then, the change of variable theorem (Eq. 3.27) becomes:

( )
( )

(3.29)
or equivalently,

( ) ( )
(3.30)
where is any arbitrary realization in the sample space of .
Let us now proceed to discuss the differences between discrete and continuous numerical data
elements.

24/09/2024 ForsChem Research Reports Vol. 9, 2024-11

10.13140/RG.2.2.22772.36485 www.forschem.org / t.me/forschem (13 / 22)
 Randomistic Data Elements
Hugo Hernandez
ForsChem Research
[email protected]

3.3. Discrete vs. Continuous Data Elements

Numerical data elements are commonly classified into discrete and continuous. The frontier
separating both types of numerical data elements is actually fuzzy, because numerical
continuity is a mathematical idealization which is never observed in the real world [7]. Formally,
we may consider the following definitions:

Discrete Data Element: Data element with a finite sample space.

Continuous Data Element: Data element with an infinite sample space.

In practice, infinite sample spaces are unfeasible. However, a discrete data element can be
assumed continuous, from a pragmatic perspective, when the size of the sample space is large
enough. The key question here is what is “large enough”. Let us assume that there is a certain
limiting size separating discrete ( ( ) ) from continuous ( ( ) ) data elements.
In principle, any arbitrary value of can be chosen from practical considerations.
Now, let us assume for simplicity that the realization values in the sample space are equally
separated by a constant distance . Even if the distance between realizations is not constant,
it is always possible to determine a minimum common separation distance , and consider
additional realizations in the sample space every , having zero probability of occurrence. This
means that the sample space size can be expressed as follows:

( )

(3.31)
This also means that there is a critical separation distance where a discrete data element
would become continuous if smaller distances are considered:

(3.32)
Notice that as . Also notice that for large values of , we may assume that any
can be approximately represented by a differential instead of a finite difference.
Now, according to the definition of the cumulative probability (Eq. 3.1) we may conclude that,
in general:

( ) ( ) ( ) ( )
(3.33)
where represents a backward finite difference.
Let us now introduce a new definition, the finite probability density function, as follows:

24/09/2024 ForsChem Research Reports Vol. 9, 2024-11

10.13140/RG.2.2.22772.36485 www.forschem.org / t.me/forschem (14 / 22)
 Randomistic Data Elements
Hugo Hernandez
ForsChem Research
[email protected]

Finite Probability Density Function: Represents the rate of change in cumulative probability per
unit change in the realization value of the data element. It is denoted by .

( )
( )

(3.34)
This definition implies that:
( ) ( )
(3.35)
For , the finite probability density function approaches a continuous derivative, simply
denoted as probability density function ( ( ) ).

Probability Density Function: Represents the derivative of the cumulative probability with respect
to the realization value of the data element. It is denoted by .

( )
( )

(3.36)
From this new definition we can obtain the following relations:
( ) ( )

(3.37)
( ) ( )

(3.38)
Thus, continuous randomistic data elements can be regarded as a particular case of the
discrete randomistic data elements, which occurs when the separation distance between
consecutive realization values (e.g. due to measurement resolution limitations) is infinitesimally
small, and approaches zero.
An important implication of the continuous approximation is that the sums considered in the
previously introduced functions and operators, can be replaced by integrals. For example, for
the expected value operator we obtain (considering Eq. 3.4 and 3.38):

( ) ∫ ( )

(3.39)
Also, any valid probability density function must necessarily satisfy the following condition
(from Eq. 2.5)

∫ ( )

(3.40)

24/09/2024 ForsChem Research Reports Vol. 9, 2024-11

10.13140/RG.2.2.22772.36485 www.forschem.org / t.me/forschem (15 / 22)
 Randomistic Data Elements
Hugo Hernandez
ForsChem Research
[email protected]

By expanding the sample space, considering differential separations between realization values
even when their probability of occurrence is zero, it is possible to describe any discrete variable
in terms of continuous probability density functions. Let us consider for example the standard
deterministic data element , defined by Eq. (3.18). The probability density function of would
be (from Eq. 3.18 and 3.38):

( ) ( )
( ) ( ) {

(3.41)
where represents Dirac’s delta function [8].
Similarly, any discrete probability distribution function can be represented by a continuous
probability density function as follows:
( )
( )
( ) ( ) ( ) ∑ ( ) ( ) ( ( ))

(3.42)

3.4. Real vs. Imaginary vs. Complex Data Elements

Numerical values can also be classified into real, imaginary and complex values. Even when
probabilities are exclusively real numbers in the range [ ] (Eq. 2.2), probability density
function values are influenced by the nature of the realization values considered.
In the case of real realization values, considering Eq. (2.2) and (3.38), we may conclude that:

( )

(3.43)
and since differential values are always considered positive, we can conclude that probability
density values of real data elements are always non-negative.
However, for imaginary and complex data elements, that is not the case [9]. For imaginary data
elements, the probability density value is always a negative imaginary number:

( )
(3.44)
and interestingly, while the variance of real data elements is always non-negative, the variance
of imaginary data elements is always non-positive.

24/09/2024 ForsChem Research Reports Vol. 9, 2024-11

10.13140/RG.2.2.22772.36485 www.forschem.org / t.me/forschem (16 / 22)
 Randomistic Data Elements
Hugo Hernandez
ForsChem Research
[email protected]

In the more general case of complex data elements, probability density values are also complex,
with a non-negative real part and a non-positive imaginary part. The variance obtained for
complex data elements is also complex.
Standard random data elements can also be defined for imaginary and complex numbers. In
general for complex numbers, the real and imaginary parts are treated independently as
elements in a vector, and are then transformed using element-wise Hadamard operators [9].
Considering that imaginary and complex data elements are seldom analyzed in practice, no
further details will be presented here.

4. Categorical Randomistic Data Elements

Unlike numerical data elements, mathematical operations cannot be directly used on

categorical data elements. It is possible, however, to indirectly use mathematical operations
when the categories can be quantified in some way [10].
The most direct approach (presented in Section 4.1) is transforming logical values
(FALSE/TRUE), into binary numerical values ( / ), usually denoted as dummy terms. But also,
categories can be represented by arbitrary numerical values, as shown in Section 4.2. The
description of purely categorical randomistic data elements is discussed in Section 4.3.

4.1. Binary Functions

Logical operators (whose outcome is a logical value: FALSE or TRUE) can be employed in both
categorical and numerical data elements, and thus, binary functions can be used to describe
any arbitrary data element.
In general, it is possible to define the following logical operator to each possible realization
value in the data element:

( ) {
(4.1)
Now, if the logical values FALSE and TRUE are arbitrary replaced by the numbers and ,
respectively, then we obtain:

( ) {
(4.2)
where ( ) is a binary function, and the realization value can be either quantitative
(numerical) or qualitative (categorical).

24/09/2024 ForsChem Research Reports Vol. 9, 2024-11

10.13140/RG.2.2.22772.36485 www.forschem.org / t.me/forschem (17 / 22)
 Randomistic Data Elements
Hugo Hernandez
ForsChem Research
[email protected]

Since the outcome of the binary function is numerical, all properties of numerical data
elements can now be employed. For example, the expected value of the binary function will
be:
( )

( (
∑ ( ) ( ) ( )
))

(4.3)
On the other hand, the variance of the binary function will be6:
( )

( (
∑ ( ) ( ) ( ) ( ) ( )
)) ( ( )
) ( ( ))

( )( ( ))
(4.4)
Also notice that ( ) represents a type III standard random data element, since
( )
and ( ( )
, with probability density function:
( ) )

( ( ) ( )( ( )) ( ) ( )
)

(4.5)
Of course, there are some limitations to this approach. Notice that we may determine the
expected value for an individual category, but we cannot determine the expected value of the
whole categorical data element (i.e. ( ) ), since is not numerical. This limitation will be
somehow circumvented using the next approach.

4.2. Quantifiable Categorical Data Elements

In some cases, the categories of a qualitative data element can be directly associated to a
quantity. This quantity may reflect the intensity of a certain property of the category, or simply
a reasonable order or hierarchy of the categories.
For example, if we have a categorical data element associated to thermal sensation, whose
space state is the set { }, then clearly we might associate a
certain representative temperature to each category (e.g. { }) or simply a
hierarchy of “hotness” (i.e. { }). Of course, these numerical assignments are arbitrary
and subjective.
This quantitative association can be represented by the quantitative function :

6
Considering that ( ) ( ).

24/09/2024 ForsChem Research Reports Vol. 9, 2024-11

10.13140/RG.2.2.22772.36485 www.forschem.org / t.me/forschem (18 / 22)
 Randomistic Data Elements
Hugo Hernandez
ForsChem Research
[email protected]

( )
(4.6)
where is a numerical data element and is a categorical data element.
For the thermal sensation example considered we have two possible quantitative functions,
associated to the intensity of the categories ( ( ) ) or to their hierarchies ( ( ) ):

( )

{
(4.7)

( )

{
(4.8)
The numerical outcomes of these functions can be treated as any arbitrary numerical data
element.
Thus, the expected value of a categorical data element ( ( )) does not exist, but it can be
alternatively represented by the expected value of an equivalent quantitative function ( ( ( ))).

5. Invariants, Variables, and Events

Previously, we have considered that data elements can be variable or invariant. This is true for
both quantitative and qualitative data elements, as well as for deterministic or random data
elements.
Unfortunately, the concept of invariance is also quite subjective. An absolute, universally
invariant data element with no changes in time, space, or any other dimension is seldom found.
Even invariant ratios such as , which could be consider as an absolute invariant, have changed
its value over time. But it is not its true physical or geometrical value which has changed, but
rather it is the value perceived and officially accepted by humanity which has changed [11].
However, if we limit time, space or any other dimension (observation condition), we may
obtain that the value of a data element always remain unchanged. So, the concept of
invariance strongly depends on the specific constraints imposed on observation conditions.
This situation is illustrated in Figure 1. The value of an arbitrary data element (ordinate) is
observed under different observation conditions (abscissa). Only one dimension of observation
conditions is considered for a simpler graphical representation. However, multiple dimensions

24/09/2024 ForsChem Research Reports Vol. 9, 2024-11

10.13140/RG.2.2.22772.36485 www.forschem.org / t.me/forschem (19 / 22)
 Randomistic Data Elements
Hugo Hernandez
ForsChem Research
[email protected]

of observation conditions are possible [12]. The observation dimension is assumed continuous,
but it can also be discrete. The blue line denotes the value of the data element observed at
each observation condition. Sudden changes in the behavior of the data element are denoted
as events.

Figure 1. Behavior of a data element value under different observation conditions. Blue line: Data
element value. Green box: Scenario 1. Yellow box: Scenario 2.

Two scenarios are considered. Each scenario represents a specific constraint on the
observation conditions. In the case of scenario 1 (green box), the data element value remains
constant, and thus, it is considered an invariant. However, for scenario 2 (yellow box), the
value of the same data element changes and is thus, a variable.
In general, a randomistic data element may remain invariant over certain ranges of observation
conditions, but may change over other ranges. Notice that the invariant range is always
delimited by two consecutive events. The span of the invariant range (between consecutive
events) is the length of the initial event.
Notice that the inclined line observed in scenario 2 might also be considered as a sequence of
many short-length events (ladder-like) if the observation condition is discrete.
Also notice that the sample space and the probability distribution function of the randomistic
data element may change depending on the constraints in observation conditions, which
ultimately define the population of the data element. For example, if we consider all possible
observations in each scenario, then for scenario 1, the sample space is limited to a single
realization value, and a Kronecker delta probability distribution function is obtained. For
scenario 2, the sample space increases. If we assume unknown the specific observation
condition within the scenario (introducing another dimension of randomness [12]), a random
probability distribution is obtained. If the observation condition is known with certainty, a
deterministic probability distribution is again obtained.

24/09/2024 ForsChem Research Reports Vol. 9, 2024-11

10.13140/RG.2.2.22772.36485 www.forschem.org / t.me/forschem (20 / 22)
 Randomistic Data Elements
Hugo Hernandez
ForsChem Research
[email protected]

6. Conclusion
Data elements represent pieces of information about a certain population. Data elements can
be classified into deterministic or random, depending on the certainty of the values in the data
element. In general, both deterministic and random data elements can be considered as a
single group denoted as randomistic data elements.
Randomistics represents the common set of rules that apply to both deterministic and random
data elements, regardless of the certainty of the information. Those rules include, for example,
the description of the data element using a probability distribution function. The concept of
randomistic data elements also applies to any data element independently of the nature of the
values involved: Quantitative/numerical or qualitative/categorical.
In the case of quantitative randomistic data elements, they can be either discrete or
continuous, or can be real, imaginary or complex. In all these cases, the same properties of
numerical randomistic data elements apply, including the expected value operator, the
moment operators, the variance operator, etc.
Randomistic data elements (either quantitative or qualitative) can also be classified into
variables or invariant, depending on the sample space of the data element for the particular
observation conditions considered. Invariant data elements will only have a single possible
realization value in the sample space. Invariants can become variables when the constraints on
the observation conditions change, and vice versa.

Acknowledgment and Disclaimer

This report provides data, information and conclusions obtained by the author(s) as a result of original
scientific research, based on the best scientific knowledge available to the author(s). The main purpose
of this publication is the open sharing of scientific knowledge. Any mistake, omission, error or inaccuracy
published, if any, is completely unintentional.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-
for-profit sectors.

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC
4.0). Anyone is free to share (copy and redistribute the material in any medium or format) or adapt
(remix, transform, and build upon the material) this work under the following terms:
 Attribution: Appropriate credit must be given, providing a link to the license, and indicating if
changes were made. This can be done in any reasonable manner, but not in any way that
suggests endorsement by the licensor.
 NonCommercial: This material may not be used for commercial purposes.

24/09/2024 ForsChem Research Reports Vol. 9, 2024-11

10.13140/RG.2.2.22772.36485 www.forschem.org / t.me/forschem (21 / 22)
 Randomistic Data Elements
Hugo Hernandez
ForsChem Research
[email protected]

References

[1] Hernandez, H. (2018). The Realm of Randomistic Variables. ForsChem Research Reports, 3, 2018-
10, 1-19. doi: 10.13140/RG.2.2.29034.16326.
[2] Hernandez, H. (2022). Standard Deterministic, Standard Random, and Randomistic Variables.
ForsChem Research Reports, 7, 2022-06, 1 - 18. doi: 10.13140/RG.2.2.36316.87688.
[3] Hernandez, H. (2023). Question Everything: Models vs. Reality. ForsChem Research Reports, 8,
2023-07, 1 - 11. doi: 10.13140/RG.2.2.35224.67845.
[4] Weisstein, E. W. (2024). Kronecker Delta. From MathWorld - A Wolfram Web Resource.
https://mathworld.wolfram.com/KroneckerDelta.html. Accessed on Sept. 17, 2024.
[5] Hernandez, H. (2018). Expected Value, Variance and Covariance of Natural Powers of
Representative Standard Random Variables. ForsChem Research Reports, 3, 2018-08, 1-19. doi:
10.13140/RG.2.2.15187.07205.
[6] Hernandez, H. (2017). Multivariate Probability Theory: Determination of Probability Density
Functions. ForsChem Research Reports, 2, 2017-13, 1-13. doi: 10.13140/RG.2.2.28214.60481.
[7] Hernandez, H. (2020). On the Discreteness of Measured Variables and the Continuous
Approximation. ForsChem Research Reports, 5, 2020-20, 1-18. doi: 10.13140/RG.2.2.27740.00646.
[8] Weisstein, E. W. (2024). Delta Function. From MathWorld - A Wolfram Web Resource.
https://mathworld.wolfram.com/DeltaFunction.html. Accessed on Sept. 20, 2024.
[9] Hernandez, H. (2019). Probability Density Functions of Imaginary and Complex Random Variables.
ForsChem Research Reports, 4, 2019-09, 1-18. doi: 10.13140/RG.2.2.20867.66083.
[10] Hernandez, H. (2021). Quantitative Analysis of Categorical Variables. ForsChem Research Reports,
6, 2021-04, 1-25. doi: 10.13140/RG.2.2.19233.33129.
[11] Berggren, L., Borwein, J., & Borwein, P. (2004). Pi: A Source Book. 3rd Edition. Springer
Science+Business Media, LLC., New York. doi: 10.1007/978-1-4757-4217-6.
[12] Hernandez, H. (2018). Multidimensional Randomness, Standard Random Variables and Variance
Algebra. ForsChem Research Reports, 3, 2018-02, 1-35. doi: 10.13140/RG.2.2.11902.48966.

24/09/2024 ForsChem Research Reports Vol. 9, 2024-11

10.13140/RG.2.2.22772.36485 www.forschem.org / t.me/forschem (22 / 22)