Loading [MathJax]/jax/output/CommonHTML/jax.js
Submit manuscript...
eISSN: 2576-4500

Aeronautics and Aerospace Open Access Journal

Research Article Volume 8 Issue 2

Novel approach to the smart constructing adequate predictive or confidence decisions for applied mathematical models under parametric uncertainty via pivotal quantities and ancillary statistics

Nicholas Nechval,1 Gundars Berzins,1 Konstantin Nechval2

1BVEF Research Institute, University of Latvia, Latvia
2Aviation Department, Riga Aeronautical Institute, Latvia

Correspondence: Nicholas Nechval, BVEF Research Institute, University of Latvia, Riga LV-1586, Latvia

Received: March 30, 2024 | Published: April 11, 2024

Citation: Nechval N, Berzins G, Nechval K. Novel approach to the smart constructing adequate predictive or confidence decisions for applied mathematical models under parametric uncertainty via pivotal quantities and ancillary statistics. Aeron Aero Open Access J. 2024;8(2):58-76. DOI: 10.15406/aaoaj.2024.08.00194

Download PDF

Abstract

The technique used here emphasizes pivotal quantities and ancillary statistics relevant for obtaining statistical predictive or confidence decisions for anticipated outcomes of applied stochastic models under parametric uncertainty and is applicable whenever the statistical problem is invariant under a group of transformations that acts transitively on the parameter space. It does not require the construction of any tables and is applicable whether the experimental data are complete or Type II censored. The proposed technique is based on a probability transformation and pivotal quantity averaging to solve real-life problems in all areas including engineering, science, industry, automation & robotics, business & finance, medicine and biomedicine. It is conceptually simple and easy to use.

Keywords: anticipated outcomes, parametric uncertainty, unknown (nuisance) parameters, elimination, pivotal quantities, ancillary statistics, new-sample prediction, within-sample prediction

Introduction

Statistical predictive or confidence decisions (under parametric uncertainty) for future random quantities (future outcomes, order statistics, etc.) based on the past and current data is the most prevalent form of statistical inference. Predictive inferences for future random quantities are widely used in risk management, finance, insurance, economics, hydrology, material sciences, telecommunications, and many other industries. Predictive inferences (predictive distributions, prediction or tolerance limits (or intervals), confidence limits (or intervals) for future random quantities on the basis of the past and present knowledge represent a fundamental problem of statistics, arising in many contexts and producing varied solutions. The approach used here is a special case of more general considerations applicable whenever the statistical problem is invariant under a group of transformations, which acts transitively on the parameter space.1–29

  1. Adequate mathematical models of cumulative distribution functions of order statistics for constructing one-sided tolerance limits (or two-sided tolerance interval) in new (future) data samples under parametric uncertainty

Theorem 1: Let us assume that Y1£ … £Yn will be a new (future) random sample of n ordered observations from a known distribution with a probability density function (pdf) fρ(y),  cumulative distribution function (cdf) Fρ(y) , where ρ is the parameter (in general, vector). Then the adequate mathematical models for a cumulative probability distribution function of the kth order statistic Yk, kÎ{1, 2, …, n}, to construct one-sided γ − content tolerance limits (or two-sided tolerance interval) for Yk with confidence level β, are given as follows:

  1. Adequate Applied Mathematical Model 1 of a Cumulative Distribution Function of the kth Order Statistic Yk is given by

Fρ(yk)0fk,nk+1(r)dr=Pρ(Ykyk|n)=nj=k(nj)[Fρ(yk)]j [1Fρ(yk)]nj.   (1)

In the above case, a (γ,β) upper, one-sided γ content tolerance limit yUk  with confidence level β  can be obtained by using the following formula:

E{Pr(Fρ(yUk)0fk,nk+1(r)drγ)}=E{Pr(Pρ(YkyUk|n)γ)}=β,  (2)

where

fk,nk+1(r)=1Β(k,nk+1)rk1(1r)(nk+1)1,   0<r<1,  (3)

is the probability density function (pdf) of the beta distribution (Beta(k,nk+1))  with the shape parameters k and nk+1.

Proof: It follows from (1) that

ddykFρ(yk)0fk,nk+1(r)dr=ddykPρ(Ykyk|n).  (4)

This ends the proof.

A (γ,β) lower, one-sided γ content tolerance limit with confidence level β  can be obtained by using the following formula:

E{Pr(Pρ(Yk>yLk|n)γ)}=E{Pr(1Fμ(yLk)0fk,nk+1(u)duγ)}=β.  (5)

A (γ,β)  two-sided γ content tolerance interval with confidence level β  can be obtained by using the following formula:

[argyLk(E{Pr(Pρ(Yk>yLk|n)γ)}=β), argyUk(E{Pr(Pρ(YkyUk|n)γ)}=β)]
=[argyLk(E{Pr(Fμ(yLk)0fk,nk+1(r)dr1γ)}=β), argyUk(E{Pr(Fρ(yUk)0fk,nk+1(r)drγ)}=β)]
=[yLk, yUk].   (6)

  1. Adequate Applied Mathematical Model 2 of a Cumulative Distribution Function of the kth Order Statistic Yk is given by

11Fρ(yk)fnk+1,k(r)dr=Pρ(Ykyk|n)=nj=k(nj)[Fρ(yk)]j [1Fρ(yk)]nj.  (7)

In the above case, a (γ,β) upper, one-sided γ content tolerance limit yUk  with confidence level β  can be obtained by using the following formula:

E{Pr(11Fρ(yUk)fnk+1,k(r)drγ)=E{Pr(Pρ(YkyUk|n)γ)}}=β,  (8)

where

fnk+1,l(u)=1Β(nk+1,k)r(nk+1)1(1r)k1fk,nk+1(r),   0<r<1,  (9)

is the probability density function (pdf) of the beta distribution (Beta(nk+1,k))  with the shape parameters nk+1 and k.

Proof: It follows from (9) that

ddyk11Fρ(yk)fnk+1,k(r)dr=ddykPρ(Ykyk|n).  (10)

This ends the proof.

A (γ,β) lower, one-sided γ content tolerance limit with confidence level β  can be obtained by using the following formula:

E{Pr(Pρ(Yk>yLk|n)γ)}=E{Pr(111Fρ(yLk)fnk+1,k(r)drγ)}=β.  (11)

A (γ,β)  two-sided γ content tolerance interval with confidence level β  can be obtained by using the following formula:

[argyLk(E{Pr(Pρ(Yk>yLk|n)γ)}=β), argyUk(E{Pr(Pρ(YkyUk|n)γ)}=β)]
=[yLk, yUk].  (12)

  1. Adequate Applied Mathematical Model 3 of a Cumulative Distribution Function of the kth Order Statistic Yk is given by

nk+1kFρ(yk)1Fρ(yk)0φk,nk+1(r)dr=Pρ(Ykyk|n)=nj=k(nj)[Fρ(yk)]j [1Fρ(yk)]nj.  (13)

In the above case, a (γ,β) upper, one-sided γ content tolerance limit yUk  with confidence level β  can be obtained by using the following formula:

E{Pr(nk+1kFρ(yUk)1Fρ(yUk)0φk,nk+1(r)drγ)}=E{Pr(Pρ(YkyUk|n)γ)}=β,  (14)

where

φk,nk+1(r)=1Β(k,nk+1) [knk+1r]k1[1+knk+1r]n+1knk+1,   r(0,),  (15)

is the probability density function (pdf) of the F distribution (F(k,nk+1)) with parameters k and nk+1, which are positive integers known as the degrees of freedom for the numerator and the degrees of freedom for the denominator.

Proof: It follows from (13) that

ddyknk+1kFρ(yk)1Fρ(yk)0φk,nk+1(r)dr=ddykPρ(Ykyk|n).  (16)

This ends the proof.

A (γ,β) lower, one-sided γ content tolerance limit with confidence level β  can be obtained by using the following formula:

E{Pr(Pρ(Yk>yLk|n)γ)}=E{Pr(1nk+1kFρ(yLk)1Fρ(yLk)0φk,nk+1(r)drγ)}=β.  (17)

A (γ,β)  two-sided γ content tolerance interval with confidence level β  can be obtained by using the following formula:

[argyLk(E{Pr(Pρ(Yk>yLk|n)γ)}=β), argyUk(E{Pr(Pρ(YkyUk|n)γ)}=β)]
=[argyLk(E{Pr(nk+1kFρ(yLk)1Fρ(yLk)0φk,nk+1(r)dr1γ)}=β), argyUk(E{Pr(nk+1kFρ(yUk)1Fρ(yUk)0φk,nk+1(r)drγ)}=β)]
=[yLk, yUk].  (18)

  1. Adequate Applied Mathematical Model 4 of a Cumulative Distribution Function of the kth Order Statistic Yk is given by

knk+11Fρ(yk)Fρ(yk)φnk+1,k(r)dr=Pρ(Ykyk|n)=nj=k(nj)[Fρ(yk)]j [1Fρ(yk)]nj.  (19)

In the above case, a (γ,β) upper, one-sided γ content tolerance limit yUk  with confidence level β  can be obtained by using the following formula:

E{Pr(knk+11Fρ(yUk)Fρ(yUk)φnk+1,k(r)drγ)}=E{Pr(Pρ(YkyUk|n)γ)}=β,  (20)

where

φnk+1,k(r)=nk+1kΒ(nk+1,k)[nk+1kr]nk[1+nk+1kr]n+1,   r(0,),  (21)

is the probability density function (pdf) of the F distribution (F(nk+1,k)) with parameters nk+1 and k, which are positive integers known as the degrees of freedom for the numerator and the degrees of freedom for the denominator.

Proof: It follows from (19) that

ddykknk+11Fρ(yk)Fρ(yk)φnk+1,k(r)dr=ddykPρ(Ykyk|n).  (22)

This ends the proof.

A (γ,β) lower, one-sided γ content tolerance limit with confidence level β  can be obtained by using the following formula:

E{Pr(1knk+11Fρ(yLk)Fρ(yLk)φnk+1,k(r)drγ)}=E{Pr(Pρ(Yk>yLk|n)γ)}=β.  (23)

A (γ,β)  two-sided γ content tolerance interval with confidence level β  can be obtained by using the following formula:

[argyLk(E{Pr(Pρ(Yk>yLk|n)γ)}=β), argyUk(E{Pr(Pρ(YkyUk|n)γ)}=β)]
=[argyLk(E{Pr(knk+11Fρ(yLk)Fρ(yLk)φnk+1,k(r)dr1γ)}=β), argyUk(E{Pr(knk+11Fρ(yUk)Fρ(yUk)φnk+1,k(r)drγ)}=β)]
=[yLk, yUk].  (24)

  1. Adequate mathematical models of conditional cumulative distribution functions of order statistic for constructing one-sided tolerance limits (or two-sided tolerance interval) in new (future) data samples under parametric uncertainty

Theorem 2: Let us assume that Y1£ … £Yn will be a new (future) random sample of n ordered observations from a known distribution with a probability density function (pdf) fρ(y),  cumulative distribution function (cdf) Fρ(y) , where ρ is the parameter (in general, vector). Then the adequate mathematical models for a conditional cumulative distribution function (ccdf) of the lth order statistic Yl, lÎ{2, …, n}, to construct one-sided γ − content tolerance limits (or two-sided tolerance interval) for Yl (1 £ k < l £ n) ), given Yk=yk, with confidence level β, are determined as follows:

  1. Adequate Applied Mathematical Model 5 of a Conditional Cumulative Distribution Function of the lth Order Statistic Yl is given by

1ˉFρ(yl)ˉFρ(yk)0flk,nl+1(r)dr=Pρ(Ylyl|Yk=yk;n)=nkj=lk(nkj)[1ˉFρ(yl)ˉFρ(yk)]j[ˉFρ(yl)ˉFρ(yk)]nkj,  (25)

In the above case, a (γ,β) upper, one-sided γ content tolerance limit yUl  with confidence level β  can be obtained by using the following formula:

E{Pr(1ˉFρ(yUl)ˉFρ(yk)0flk,nl+1(r)drγ)}=E{Pr(Pρ(YlyUl|Yk=yk;n)γ)}=β,  (26)

where ˉFμ(z)=1Fμ(z),

flk,nl+1(r)=rlk1(1r)(nl+1)1Β(lk,nl+1),   0<r<1,  (27)

is the probability density function (pdf) of the beta distribution (Beta(lk,nl+1))  with shape parameters l−k and n−l+1.

Proof: It follows from (25) that

ddyl1ˉFρ(yl)ˉFρ(yk)0flk,nl+1(r)dr=ddylPρ(Ylyl|Yk=yk;n).  (28)

This ends the proof.

A (γ,β) lower, one-sided γ content tolerance limit with confidence level β  can be obtained by using the following formula:

E{Pr(11ˉFρ(yLl)ˉFρ(yk)0flk,nl+1(r)drγ)}=E{Pr(Pρ(Yl>yLl|Yk=yk;n)γ)}=β.  (29)

A (γ,β)  two-sided γ content tolerance interval with confidence level β  can be obtained by using the following formula:

[argyLl(E{Pr(Pρ(Yl>yLl|Yk=yk;n)γ)}=β), argyUl(E{Pr(Pρ(YlyUl|n)γ)}=β)]
=[argyLk(E{Pr(1ˉFρ(yLl)ˉFρ(yk)0flk,nl+1(r)dr1γ)}=β), argyUk(E{Pr(1ˉFρ(yUl)ˉFρ(yk)0flk,nl+1(r)drγ)}=β)]
=[yLl, yUl].  (30)

  1. Adequate Applied Mathematical Model 6 of a Conditional Cumulative Distribution Function of the lth Order Statistic Yl is given by

1ˉFρ(yl)ˉFρ(yk)fnl+1,lk(r)dr=Pρ(Ylyl|Yk=yk;n)=nkj=lk(nkj)[1ˉFρ(yl)ˉFρ(yk)]j[ˉFρ(yl)ˉFρ(yk)]nkj,  (31)

In the above case, a (γ,β) upper, one-sided γ content tolerance limit yUl  with confidence level β  can be obtained by using the following formula:

E{Pr(1ˉFρ(yUl)ˉFρ(yk)fnl+1,lk(r)drγ)}=E{Pr(Pρ(YlyUl|Yk=yk;n)γ)}=β,  (32)

where ˉFρ(y)=1Fρ(y),

fnl+1,lk(r)=r(nl+1)1(1r)lk1Β(nl+1,lk),   0<r<1,  (33)

is the probability density function (pdf) of the beta distribution (Beta(nl+1,lk))  with shape parameters n−l+1 and l−k.

Proof: It follows from (31) that

ddyl1ˉFρ(yl)ˉFρ(yk)fnl+1,lk(r)dr=ddylPρ(Ylyl|Yk=yk;n).  (34)

This ends the proof.

A (γ,β) lower, one-sided γ content tolerance limit with confidence level β  can be obtained by using the following formula:

E{Pr(11ˉFρ(yLl)ˉFρ(yk)fnl+1,lk(r)drγ)}=E{Pr(Pρ(Yl>yLl|Yk=yk;n)γ)}=β.  (35)

A (γ,β)  two-sided γ content tolerance interval with confidence level β  can be obtained by using the following formula:

[argyLl(E{Pr(Pρ(Yl>yLl|Yk=yk;n)γ)}=β), argyUl(E{Pr(Pρ(YlyUl|Yk=yk;n)γ)}=β)]
=[argyLl(E{Pr(1ˉFρ(yLl)ˉFρ(yk)fnl+1,lk(r)dr1γ)}=β), argyUk(E{Pr(1ˉFρ(yUl)ˉFρ(yk)fnl+1,lk(r)drγ)}=β)]
=[yLl, yUl].  (36)

This ends the proof.

  1. Adequate Applied Mathematical Model 7 of a Conditional Cumulative Distribution Function of the lth Order Statistic Yl is given by

nl+1lk(1ˉFρ(yl)ˉFρ(yk))/ˉFρ(yl)ˉFρ(yk)0flk,nl+1(r)dr=Pρ(Ylyl|Yk=yk;n)=nkj=lk(nkj)[1ˉFρ(yl)ˉFρ(yk)]j[ˉFρ(yl)ˉFρ(yk)]nkj,  (37)

In the above case, a (γ,β) upper, one-sided γ content tolerance limit yUl  with confidence level β  can be obtained by using the following formula:

E{Pr(nl+1lk(1ˉFρ(yUl)ˉFρ(yk))/ˉFρ(yUl)ˉFρ(yk)0flk,nl+1(r)drγ)}=E{Pr(Pρ(YlyUl|Yk=yk;n)γ)}=β,  (38)

where ˉFρ(y)=1Fρ(y),

flk,nl+1(r)=lknl+1Β(lk,nl+1)[lknl+1r]lk1[1+lknl+1r]nk+1, r(0,),   (39)

is the probability density function (pdf) of the F distribution (F(lk,nl+1)) with parameters lk and nl+1, which are positive integers known as the degrees of freedom for the numerator and the degrees of freedom for the denominator.

Proof: It follows from (36) that

ddylnl+1lk(1ˉFρ(yl)ˉFρ(yk))/ˉFρ(yl)ˉFρ(yk)0flk,nl+1(r)dr=ddylPρ(Ylyl|Yk=yk;n).  (40)

This ends the proof.

A (γ,β) lower, one-sided γ content tolerance limit with confidence level β  can be obtained by using the following formula:

E{Pr(1nl+1lk(1ˉFρ(yLl)ˉFρ(yk))/ˉFρ(yLl)ˉFρ(yk)0flk,nl+1(r)drγ)}=E{Pr(Pρ(Yl>yLl|Yk=yk;n)γ)}=β.  (41)

A (γ,β)  two-sided γ content tolerance interval with confidence level β  can be obtained by using the following formula:

[argyLl(E{Pr(Pρ(Yl>yLl|Yk=yk;n)γ)}=β), argyUl(E{Pr(Pr(Pρ(YlyUl|Yk=yk;n)γ))}=β)]
=[argyLl(E{Pr(nl+1lk(1ˉFρ(yLl)ˉFρ(yk))/ˉFρ(yLl)ˉFρ(yk)0flk,nl+1(r)dr1γ)}=β),argyUk(E{Pr(nl+1lk(1ˉFρ(yUl)ˉFρ(yk))/ˉFρ(yUl)ˉFρ(yk)0flk,nl+1(r)drγ)}=β)] =[yLl, yUl].  (42)

This ends the proof.

  1. Adequate Applied Mathematical Model 8 of a Conditional Cumulative Distribution Function of the lth Order Statistic Yl is given by

lknl+1ˉFρ(yl)ˉFρ(yk)/(1ˉFρ(yl)ˉFρ(yk))fnl+1,lk,(r)dr =Pρ(Ylyl|Yk=yk;n)
=nkj=lk(nkj)[1ˉFρ(yl)ˉFρ(yk)]j[ˉFρ(yl)ˉFρ(zk)]nkj  (43)

In the above case, a (γ,β) upper, one-sided γ content tolerance limit yUl  with confidence level β  can be obtained by using the following formula:

E{Pr(lknl+1ˉFρ(yUl)ˉFρ(yk)/(1ˉFρ(yUl)ˉFρ(yk))fnl+1,lk,(r)drγ)}=E{Pr(Pρ(YlyUl|Yk=yk;n)γ)}=β,   (44)

where ˉFρ(y)=1Fρ(y),

fnl+1,lk(r)=lknl+1Β(lk,nl+1)[lknl+1r]lk1[1+lknl+1r]nk+1, r(0,),   (45)

is the probability density function (pdf) of the F distribution (F(nl+1,lk)) with parameters nl+1 and lk, which are positive integers known as the degrees of freedom for the numerator and the degrees of freedom for the denominator.

Proof: It follows from (36) that

ddyllknl+1ˉFρ(yl)ˉFρ(yk)/(1ˉFρ(yl)ˉFρ(yk))fnl+1,lk,(r)dr=ddylPρ(Ylyl|Yk=yk;n).  (46)

This ends the proof.

A (γ,β) lower, one-sided γ content tolerance limit with confidence level β  can be obtained by using the following formula:

E{Pr(1lknl+1ˉFρ(yLl)ˉFρ(yk)/(1ˉFρ(yLl)ˉFρ(yk))fnl+1,lk,(r)drγ)}=E{Pr(Pρ(Yl>yLl|Yk=yk;n)γ)}=β.  (47)

A (γ,β)  two-sided γ content tolerance interval with confidence level β  can be obtained by using the following formula:

[argyLl(E{Pr(Pρ(Yl>yLl|Yk=yk;n)γ)}=β), argyUl(E{Pr(Pr(Pρ(YlyUl|Yk=yk;n)γ))}=β)]
=[argyLl(E{Pr(lknl+1ˉFρ(yLl)ˉFρ(yk)/(1ˉFρ(yLl)ˉFρ(yk))fnl+1,lk,(r)dr1γ)}=β),argyUk(E{Pr(lknl+1ˉFρ(yUl)ˉFρ(yk)/(1ˉFρ(yUl)ˉFρ(yk))fnl+1,lk,(r)drγ)}=β)] =[yLl, yUl].  (48)

This ends the proof.

  1. Two-parameter exponential distribution

Let Y = (Y1 £ ... £ Ym) be the first m ordered observations (order statistics) in a sample of size h from the two-parameter exponential distribution with the probability density function

fρ(y)=ϑ1exp(yυϑ),   ϑ>0, υ0,  (49)

and the cumulative probability distribution function

Fρ(y)=1exp(yυϑ), ˉFρ(y)=1Fρ(y)=exp(yυϑ),   (50)

where ρ=(υ,ϑ), υ is the shift parameter and ϑ is the scale parameter. It is assumed that these parameters are unknown. In Type II censoring, which is of primary interest here, the number of survivors is fixed and Y is a random variable. In this case, the likelihood function is given by

L(υ,ϑ)=mi=1fρ(yi)(ˉFρ(ym))hm =1ϑmexp([mi=1(yiυ)+(hm)(ymυ)]/ϑ)
=1ϑmexp([mi=1(yiy1+y1υ)+(hm)(ymy1+y1υ)]/ϑ)
=1ϑm1exp([mi=1(yiy1)+(hm)(ymy1)]/ϑ)
×1ϑexp(h(y1υ)ϑ) =1ϑm1exp(smϑ)×1ϑexp(h(s1υ)ϑ),  (51)

where

S=(S1=Y1, Sm=mi=1(YiY)1+(hm)(YmY1))  (52)

is the complete sufficient statistic for ρ. The probability density function of S = (S1, Sm) is given by

fρ(s1,sm) =1ϑm1exp(smϑ)×1ϑexp(h(s1υ)ϑ)1sm2m0sm2mϑm1exp(smϑ)dsm×1q0hϑexp(h(s1υ)ϑ)ds1
=1ϑm1exp(smϑ)×1ϑexp(h(s1υ)ϑ)Γ(m1)sm2m×1h
=1Γ(m1)ϑm1sm2mexp(smϑ) ×hϑexp(h(s1υ)ϑ)=fϑ(sm)fρ(s1),  (53)

where

fρ(s1)=hϑexp(h(s1υ)ϑ),   s1υ,  (54)

fϑ(sm)=1Γ(m1)ϑm1sm2mexp(smϑ),   sm0.  (55)

V1=S1υϑ   (56)

is the pivotal quantity, the probability density function of which is given by

f1(v1)=hexp(hv1),   v10,  (57)

Vm=Smϑ  (58)

is the pivotal quantity, the probability density function of which is given by

fm(vm)=1Γ(m1)vm2mexp(vm),   vm0.  (59)

  1. Constructing a (γ,β) upper, one-sided γ content tolerance limit with confidence level β for the case of Model 1

Theorem 3: Let Y1£…£Ym be the first m ordered observations from the preliminary sample of size h from a two-parameter exponential distribution defined by the probability density function (49). Then the upper one-sided γ-content tolerance limit (with a confidence level β) yUk  on the kth order statistic Yk from a set of n future ordered observations Y1£…£Yn also from the distribution (49), which satisfies

E{Pr(Pρ(YkyUk|n)γ)}=β,  (60)

is given by

yUk={S1+Smh[1(Ωhγβ)1m1],   if   (Ωhγβ)1m11,S1+Smh[(Ωhγβ)1m11],   if   (Ωhγβ)1m1>1,  (61)

where

Ωγ=1q(k,nk+1),γ(Beta(k,n-k+1), γ  quantile).  (62)

Proof: It follows from (2) and (3) that

E{Pr(Pρ(YkyUk|n)γ)}
=E{Pr(Fρ(yUk)0fk,nk+1(r)drγ)} =E{Pr(1exp(yUkυϑ)qk,nk+1;γ)}
=E{Pr(exp(yUkυϑ)1qk,nk+1;γ)}
=E{Pr(yUkυϑln(1qk,nk+1;γ))}=E{Pr(yUkυϑln(1qk,nk+1;γ))}
=E{Pr(yUkS1SmSmϑ+S1υϑln(1qk,nk+1;γ))}
=E{Pr(S1υϑyUkS1SmSmϑln(1qk,nk+1;γ))}
=E{Pr(V1ηUkVmlnΩγ)}=E{1Pr(V1ηUkVmlnΩγ)}=E{1ηUkVmlnΩγ0f1(v1)dv1},  (63)

where

  ηUk=yUkS1Sm.  (64)

It follows from (63) and (64) that

E{1ηUkVmlnΩγ0f1(v1)dv1} =E{1ηUkVmlnΩγ0hexp(hv1)dv1}
=E{1[1exp(h[ηUkVmlnΩγ])]} =E{exp(hηUkVm)exp(lnΩhγ)}
=E{Ωhγexp(hηUkVm)} =0(Ωhγexp(hηUkvm))fm(vm)dvm
=0(Ωhγexp(hηUkvm))1Γ(m1)vm2mexp(vm)dvm=Ωhγ01Γ(m1)vm2mexp(vm[1hηUk])dvm
=Ωhγ[1hηUk]m1=β.  (65)

It follows from (64) and (65) that

ηUk=yUkS1Sm=1h(1[Ωhγβ]1m1).  (66)

It follows from (66) that

yUk=S1+Smh(1[Ωhγβ]1m1).  (67)

Then (61) follows from (67), this ends the proof.

  1. Constructing a (γ,β) lower, one-sided γ content tolerance limit with confidence level β for the case of Model 1

Theorem 4: Let Y1£…£Ym be the first m ordered observations from the preliminary sample of size h from a two-parameter exponential distribution defined by the probability density function (49). Then the lower one-sided γ-content tolerance limit (with a confidence level β) yLk on the kth order statistic Yk from a set of n future ordered observations Y1£…£Yn also from the distribution (49)), which satisfies

E{Pr(Pμ(Yk>yLk|n)γ)}=β,  (68)

is given by

yLk={S1+Smh[1(Ωh1γ1β)1m1],   if   (Ωh1γ1β)1m11,S1+Smh[(Ωh1γ1β)1m11],   if   (Ωh1γ1β)1m1>1,  (69)

where

Ω1γ=1 q(k,nk+1),1γ(Beta(k,n-k+1), 1γ  quantile).  (70)

Proof: It follows from (3) and (5) that

E{Pr(Pρ(Yk>yLk|n)γ)} =E{Pr(Fρ(yLk)0fk,nk+1(r)dr1γ)}
=E{Pr(exp(yLkυϑ)1qk,nk+1;1γ)}
=E{Pr(yLkS1SmSmϑ+S1υϑln(1qk,nk+1;1γ))}
=E{Pr(S1υϑyLkS1SmSmϑln(1qk,nk+1;1γ))}
=E{Pr(V1ηLkVmlnΩ1γ)}=E{ηLkVmlnΩ1γ0f1(v1)dv1},  (71)

where

  ηLk=yLkS1Sm.  (72)

It follows from (57) and (71) that

E{ηLkVmlnΩ1γ0f1(v1)dv1}=E{ηLkVmlnΩ1γ0hexp(hv1)dv1}
=E{1exp(h[ηLkVmlnΩ1γ])} =E{1exp(hηLkVm)exp(qlnΩ1γ)}
=E{1Ωh1γexp(hηLkVm)} =0(1Ωh1γexp(hηLkvm))fm(vm)dvm
=0(1Ωh1γexp(hηLkvm))1Γ(m1)vm2mexp(vm)dvm=1Ωh1γ01Γ(m1)vm2mexp(vm[1hηLk])dvm
=1Ωh1γ[1hηLk]m1=β.  (73)

It follows from (72) and (73) that

ηLk=yLkS1Sm=1h(1[Ωh1γ1β]1m1).  (74)

It follows from (74) that

yLk=S1+Smh(1[Ωh1γ1β]1m1).  (75)

Then (69) follows from (75), this ends the proof.

  1. Numerical Practical Example

Let us assume that k =5, m =8, h =10, n=12, γ = β = 0.95,

S=(S1=Y1=9, Sm=mi=1(YiY)1+(hm)(YmY1))  

=(S1=9, Sm=0+1+2+4+6+10+15+23+(108)23=107),  (76)

Then , the (γ=0.95,β=0.95) upper, one-sided γ content tolerance limit yUk  with confidence level β  can be obtained from (61), where the quantile of Beta(k,n-k+1),γ  is given by

q(k,nk+1),γ=0.609138,  (77)

Ω1γ=1q(k,nk+1),1γ=10.609138=0.390862.  (78)

It follows from (61), (76) and (78) that

yUk=S1+Smh[1(Ωhγβ)1m1]=9+10710[(1[0.390862]100.95)181]=9+7.883285=16.883285.  (79)

The (γ=0.95,β=0.95) lower, one-sided γ content tolerance limit yUk  with confidence level β  can be obtained from (69), where the quantile of Beta(k,n-k+1),1γ  is given by

q(k,nk+1),1γ=0.181025,  (80)

Ω1γ=1q(k,nk+1),1γ=10.181025=0.818975.  (81)

It follows from (69), (76) and (81) that

yLk=S1+Smh[(Ωhγ1β)1m11]=9+10710[([0.818975]1010.95)1811]=9+10710[1.153353261]=10.64088.  (82)

The (γ=0.95,β=0.95)  two-sided γ content tolerance interval with confidence level β  can be obtained by using (6), (79) and (82):

[yLk, yUk]=[10.64088, 16.883285].  (83)

  1. New intelligent transformation technique for derivation of the density function of the student’s T distribution

Theorem 5: If W1N(0,1)  and W2χ2(υ)  are independent random variables, then

W1/W2/υ=T(υ),  (84)

where t(υ)   follows the student’s t distribution with υ  degrees of freedom,

t(υ)f(t)=Γ((υ+1)/2))πυ Γ(υ/2)[1+t2υ](υ+1)/2,  <t<.  (85)

Proof.

w1f1(w1)=12πexp(w212),   <w1<,   (86)

where

w1=t[w2υ]1/2,   dw1=[w2υ]1/2dt.  (87)

It follows from (86) and (87) that

f1(w1)dw1=12πexp(w212)dw1
=12πexp(t2[w2/υ]2)[w2υ]1/2dt=f(t|w2)dt,  <t<.    (88)

w2f2(w2)=1Γ(υ/2)2υ/2w(υ/2)12exp(w22),   0<w2<.  (89)

It follows from (88) and (89) that

f(t)=0f(t|w2)f2(w2)dw2
=012πexp(t2[w2/υ]2)[w2υ]1/21Γ(υ/2)2υ/2w(υ/2)12exp(w22)dw2
=01πυΓ(υ/2)2(υ+1)/2w(υ+1)/2)12exp(w22[1+t2υ])dw2
=Γ((υ+1)/2))πυ Γ(υ/2)[1+t2υ](υ+1)/2,  <t<.  (90)

This ends the proof.

  1. Confidence interval for the difference of means of two different normal populations

In most applications, two populations are compared using the difference in the means. Let U1, U2, ..., Um be a sample of size m from a normal population having mean μm  and variance σ2m  and let Z1, ..., Zn be a sample of size n from a different normal population having mean μn  and variance σ2n  and suppose that the two samples are independent of each other. We are interested in constructing a confidence interval for μmμn.  To obtain this confidence interval, we need the distribution of ˉUmˉZn, where

ˉUm=mi=1Ui/mN(μm,σ2m/m),   ˉZn=mi=1Zi/nN(μn,σ2n/n)  (91)

It follows from (91) that

ˉUmˉZnN(μmμn,σ2mm+σ2nn).  (92)

It follows from (92) that

ˉUmˉZn(μmμn)σ2m/m+σ2n/n=W1N(0,1).   (93)

This is independent of

mi=1(UiˉUm)2/σ2m=(m1)σ2mmi=1(UiˉUm)2(m1)=(m1)S2mσ2mχ2m1  (94)

and

ni=1(ZiˉZn)2/σ2n=(n1)σ2nni=1(ZiZn)2(n1)=(n1)S2nσ2nχ2n1,  (95)

where

(m1)S2mσ2m+(n1)S2nσ2n=W2χ2(m+n2).  (96)

Taking (84), (93) and (96) into account, we have that

W1W2/(m+n2)=ˉUmˉZn(μmμn)σ2m/m+σ2n/n[(m1)S2mσ2m+(n1)S2nσ2n]/(m+n2)
=ˉUmˉZn(μmμn)(m1)S2m/σ2m+(n1)S2n/σ2nm+n2σ2m/m+σ2n/n=T(m+n2)f(t),  (97)

where T(m+n-2) is a t-random variable with m + n – 2 degrees of freedom,

f(t)=Γ((m+n1)/2)π(m+n2) Γ((m+n2)/2)[1+t2m+n2](m+n1)/2,  <t<.  (98)

Using (97) and (98), it can be obtained a 100(1-a)% confidence interval for ˉUmˉZn(μmμn)  from

P(t1T(m+n2)t2)
=P(t1ˉUmˉZn(μmμn)(m1)S2m/σ2m+(n1)S2n/σ2nm+n2σ2m/m+σ2n/nt2)
=P(t1(m1)S2m/σ2m+(n1)S2n/σ2nm+n2σ2m/m+σ2n/nˉUmˉZn(μmμn)                t2(m1)S2m/σ2m+(n1)S2n/σ2nm+n2σ2m/m+σ2n/n)=1α  (99)

by suitably choosing the decision variables t1 and t2. Hence, the statistical confidence interval for ˉUmˉZn(μmμn)  is given by

[t1(m1)S2m/σ2m+(n1)S2n/σ2nm+n21σ2m/m+σ2n/n, t2(m1)S2m/σ2m+(n1)S2n/σ2nm+n21σ2m/m+σ2n/n].  (100)

The length of the statistical confidence interval for ˉUmˉZn(μmμn)  is given by

L(t1,t2|(m1)S2m/σ2m+(n1)S2n/σ2nm+n2σ2m/m+σ2n/n)
=(t2t1)(m1)S2m/σ2m+(n1)S2n/σ2nm+n2σ2m/m+σ2n/n.  (101)

In order to find the confidence interval of shortest-length for ˉUmˉZn(μmμn) , we should find a pair of decision variables t1 and t2 such that (101) is minimum.

It follows from (98) and (99) that

t2t1f(t)dt=t20f(t)dtt10f(t)dt =(1α+p)p=1α,  (102)

where p (0pα)  is a decision variable,

t20f(t)dt=1α+p  (103)

and

t10f(t)dt=p.  (104)

Then t2 represents the (1α+p) - quantile, which is given by

t2=q1α+p;(t(m+n2)),  (105)

t1represents the p - quantile, which is given by

t1=qp;(t(m+n2)).  (106)

The shortest length confidence interval for ˉUmˉZn(μmμn)  can be found as follows:

Minimize

(t2t1)2=(q1α+p;(t(m+n2))qp;(t(m+n2)))2   (107)

subject to

0pα,  (108)

The optimal numerical solution minimizing (t2t1)2 can be obtained using the standard computer software "Solver" of Excel 2016. If σ2m=σ2n,  it follows from (101) that

L(t1,t2|(m1)S2m+(n1)S2nm+n2m+nmn)=(t2t1)(m1)S2m+(n1)S2nm+n2m+nmn.  (109)

If, for example, m=58, n=27, a = 0.05, ˉUm=70.7,   ˉZn=76.13,   S2m=(1.8)2, S2n=(2.42)2, then the optimal numerical solution of (107) is given by

p=0.025,   t1=qp;(t(m+n2))=1.98896, t2=q1α+p;(t(m+n2))=1.98896   (110)

and it follows from (99) and (109) that the 100(1-a)% confidence interval of shortest-length (or equal tails) for μ1μ2 is given by

(μmμn)((ˉUmˉZn)t2(m1)S2m+(n1)S2nm+n2m+nmn,(ˉUmˉZn)t1(m1)S2m+(n1)S2nm+n2m+nmn)=(6.330947,4.52905)   (111)

or

6.330947  μmμn4.52905.  (112)

  1. Confidence interval for the ratio of means of two different normal populations

Ratio in the means is used to compare two populations of positive data. Let U1, U2, ..., Um be a sample of size m from a normal population having mean μm  and variance σ2m  and let U1, ..., Un be a sample of size n from a different normal population having mean μn  and variance σ2n  and suppose that the two samples are independent of each other. We are interested in constructing a confidence interval for the ratio of means (μm,μn) of two different normal populations To obtain this confidence interval, we need the distribution of ˉUmκˉUn, where

ˉUm=mi=1Ui/mN(μm,σ2m/m),   ˉUn=ni=1Ui/nN(μn,σ2n/n)  (113)

It can be shown that

ˉUmκˉUnN(μmκμn,σ2mm+κ2σ2nn)  (114)

or

ˉUmκˉUn(μmκμn)σ2mm+κ2σ2nn=W1N(0,1).  (115)

This is independent of

mi=1(UiˉUm)2/σ2m=(m1)σ2mmi=1(UiˉUm)2(m1)=(m1)S2mσ2mχ2m1  (116)

and

nj=1(UjˉUn)2/σ2n=(n1)σ2nnj=1(UjˉUn)2(n1)=(n1)S2nσ2nχ2n1,  (117)

where

(m1)S2mσ2m+(n1)S2nσ2n=W2χ2(m+n2).  (118)

It follows from (84), (115) and (118) that

W1W2/(m+n2)=ˉUmκˉUn(μmκμn)σ2mm+κ2σ2nn1[(m1)S2mσ2m+(n1)S2nσ2n]/(m+n2)  

=ˉUmκˉUn(μmκμn)(m1)S2m/σ2m+(n1)S2n/σ2nm+n2σ2m/m+κ2σ2n/n=T(m+n2)f(t),  (119)

where T(m+n-2) is a t-random variable with m + n – 2 degrees of freedom. Taking Theorem 5 into account, we have that

f(t)=Γ((m+n1)/2)π(m+n2) Γ((m+n2)/2)[1+t2m+n2](m+n1)/2,  <t<.  (120)

Using (119) and (120), it can be obtained a 100(1-a)% confidence interval for ˉUmκˉUn(μmκμn)  from

P(t1T(m+n2|ˉUmκˉUn(μmκμn))t2)
=P(t1(m1)S2m/σ2m+(n1)S2n/σ2nm+n2σ2m/m+κ2σ2n/nˉUmκˉUn(μmκμn)                  t2(m1)S2m/σ2m+(n1)S2n/σ2nm+n2σ2m/m+κ2σ2n/n)=1α  (121)

By suitably choosing the decision variables t1 and t2. Hence, the statistical confidence interval for ˉUmκˉUn(μmκμn)  is given by

[t1(m1)S2m/σ2m+(n1)S2n/σ2nm+n21σ2m/m+κ2σ2n/n, t2(m1)S2m/σ2m+(n1)S2n/σ2nm+n21σ2m/m+κ2σ2n/n].  (122)

The length of the statistical confidence interval for ˉUmκˉUn(μmκμn)  is given by

L(t1,t2|(m1)S2m/σ2m+(n1)S2n/σ2nm+n2σ2m/m+κ2σ2n/n) =(t2t1)(m1)S2m/σ2m+(n1)S2n/σ2nm+n2σ2m/m+κ2σ2n/n.  (123)

In order to find the confidence interval of shortest-length for ˉUmκˉUn(μmκμn) , we should find a pair of decision variables t1 and t2 such that (123) is minimum. It follows from (121) and (123) that

t2t1f(t)dt=t20f(t)dtt10f(t)dt =(1α+p)p=1α,  (124)

where p (0pα)  is a decision variable,

t20f(t)dt=1α+p  (125)

and

t10f(t)dt=p.  (126)

Then t2 represents the (1α+p) - quantile, which is given by

t2=q1α+p;(t(m+n2)),  (127)

t1 represents the p  - quantile, which is given by

t1=qp;(t(m+n2)).  (128)

The shortest length confidence interval for ˉUmκˉUn(μmκμn)  can be found as follows:

Minimize

(t2t1)2=(q1α+p;(t(m+n2))qp;(t(m+n2)))2   (129)

subject to

0pα,  (130)

The optimal numerical solution minimizing (t2t1)2 can be obtained using the standard computer software "Solver" of Excel 2016. If σ2m=σ2n,  it follows from (123) that

L(t1,t2|(m1)S2m+(n1)S2nm+n21m+κ2n)=(t2t1)(m1)S2m+(n1)S2nm+n21m+κ2n.  (131)

If, for example, m=6, n=4, a = 0.05,ˉUm=117.5,   ˉUn=126.8,   S2m=(9.7)2, S2n=(12)2, then the optimal numerical solution of (129) is given by

p=0.025,   t1=qp;(t(m+n2))=2.306,   t2=q1α+p;(t(m+n2))=2.306  (132)

and it follows from (121) and (131) that the 100(1-a)% confidence interval of shortest-length (or equal tails) for μ1κμ2  is given by

(ˉUmκˉUn(μmκμn)t1(m1)S2m+(n1)S2nm+n21m+κ2n,ˉUmκˉUn(μmκμn)t2(m1)S2m+(n1)S2nm+n21m+κ2n)  (133)

If  it follows from (133) that

(μmμn)((ˉUmˉUn)t2(m1)S2m+(n1)S2nm+n21m+1n,(ˉUmˉUn)t1(m1)S2m+(n1)S2nm+n21m+1n)
=((117.5126.8)2.306×10.616+14,(117.5126.8)+2.306×10.616+14)=(25.07, 6.47)  (134)

or

25.07 < μmμn<6.47.  (135)

An analytical expression for determining the optimal value of κ  (the ratio in means of two different normal populations) can be obtained from (121), where it is assumed that σ2m=σ2n  and (μmκμn)=0:  

(t1(m1)S2m+(n1)S2nm+n21/m+κ2/nˉUmκˉUnt2(m1)S2m+(n1)S2nm+n21/m+κ2/n)
=(κˉUn+t1(m1)S2m+(n1)S2nm+n21/m+κ2/nˉUm,ˉUmκˉUn+t2(m1)S2m+(n1)S2nm+n21/m+κ2/n)
=(κ+t1(m1)S2m+(n1)S2nm+n2ˉUn1/m+κ2/nˉUmˉUn,ˉUmˉUnκ+t2(m1)S2m+(n1)S2nm+n21/m+κ2/n)
=(κˉUmˉUnt1(m1)S2m+(n1)S2nm+n2ˉUn1/m+κ2/n,κˉUmˉUnt2(m1)S2m+(n1)S2nm+n2ˉUn1/m+κ2/n)
=(κ0.926656+2.30610.6126.81/6+κ2/4,κ0.9266562.30610.6126.81/6+κ2/4)
=(κ0.926656+0.1927730.166667+0.25κ2,κ0.9266560.1927730.166667+0.25κ2)
(                                minimize:(κ0.9266560.1927730.166667+0.25κ2)2,(κ0.926656+0.1927730.166667+0.25κ2)2,                           subject to: κ0.)=(κ1.05526,κ0.815431).  (136)

Thus, it follows from (136) that

κ(0.815431, 1.05526).  (137)

Conclusion

The new intelligent computational models proposed in this paper are conceptually simple, efficient, and useful for constructing accurate statistical tolerance or prediction limits and shortest-length or equal-tailed confidence intervals under the parametric uncertainty of applied stochastic models. The methods listed above are based on adequate computational models of the cumulative distribution function of order statistics and constructive use of the invariance principle in mathematical statistics. These methods can be used to solve real-life problems in all areas including engineering, science, industry, automation & robotics, machine learning, business & finance, medicine and biomedicine, optimization, planning and scheduling.

Acknowledgments

None.

Conflicts of interest

The authors declare that there is no conflict of interest.

References

  1. Nechval NA, Vasermanis EK. Improved decisions in statistics. Riga: Izglitibas soli; 2004.
  2. Nechval NA, Berzins G, Purgailis M, et al. Improved estimation of state of stochastic systems via invariant embedding technique. WSEAS Transactions on Mathematics. 2008;7:141–159.
  3. Nechval NA, Nechval KN, Danovich V, et al. Optimization of new-sample and within-sample prediction intervals for order statistics. Proceedings of the 2011 World Congress in Computer Science, Computer Engineering, and Applied Computing, WORLDCOMP'11; 2011 July 18-21; Las Vegas Nevada, USA: CSREA Press; 2011. p. 91–97.
  4. Nechval NA, Nechval KN, Berzins G. A new technique for intelligent constructing exact g-content tolerance limits with expected (1 − a)-confidence on future outcomes in the Weibull case using complete or Type II censored data. AC&CS. 2018;52:476–488.
  5. Nechval NA, Berzins G, Nechval KN. Intelligent technique of constructing exact statistical tolerance limits to predict future outcomes under parametric uncertainty for prognostics and health management of complex systems. IJCSIA. 2019;9:30–47.
  6. Nechval NA, Berzins G, Nechval KN, et al. A new technique of intelligent constructing unbiased prediction limits on future order statistics coming from an inverse Gaussian distribution under parametric uncertainty. AC&CS. 2019;53:223–235.
  7. Nechval NA, Berzins G, Nechval KN. A novel intelligent technique for product acceptance process optimization on the basis of misclassification probability in the case of log-location-scale distributions. In: Wotawa F, editors. Advances and Trends in Artificial Intelligence From Theory to Practice IEA/AIE 2019: Lecture Notes in Computer Science; Switzerland: Springer Nature; 2019;11606:801–818.
  8. Nechval NA, Berzins G, Nechval KN. A novel intelligent technique of invariant statistical embedding and averaging via pivotal quantities for optimization or improvement of statistical decision rules under parametric uncertainty. WSEAS Transactions on Mathematics. 2020;19:17–38.
  9. Nechval NA, Berzins G, Nechval KN, et al. A novel technique for optimization of statistical decisions under parametric uncertainty through invariant statistical embedding and averaging in terms of pivotal quantities. Journal of Physics: Conference Series. 2020;1603:012022.
  10. Nechval NA, Berzins G, Nechval KN, et al. Intelligent constructing optimal airline seat protection levels for multiple nested fare classes of single-leg flights. Journal of Physics: Conference Series. 2020;1603:012023.
  11. Nechval NA, Berzins G, Nechval KN. A new technique of invariant statistical embedding and averaging via pivotal quantities for intelligent constructing efficient statistical decisions under parametric uncertainty. AC&CS. 2020:54:191–206.
  12. Nechval NA, Berzins G, Nechval KN, et al. A new unified computational approach to intelligent constructing shortest-length or equal tails confidence intervals under parametric uncertainty. Journal of Applied Mathematics and Computation. 2020:206–223.
  13. Nechval NA, Berzins G, Nechval KN. Cost-effective planning reliability-based inspections of fatigued structures in the case of log-location-scale distributions of lifetime under parametric uncertainty. In: Piero Baraldi, Francesco Di Maio, Enrico Zio, editors. ESREL2020-PSAM15: Proceedings of the 30th European Safety and Reliability Conference and the 15th Probabilistic Safety Assessment and Management Conference; 2020 November 1-6; Venice, Italy: 2020. p. 455–462.
  14. Nechval NA. Intelligent constructing exact tolerance limits for prediction of future outcomes under parametric uncertainty. In: Encyclopedia of Information Science and Technology. 5th ed. USA: IGI Global; 2020;701–729.
  15. Nechval NA, Berzins G, Nechval KN, et al. A novel unified computational approach to constructing shortest-length or equal tails confidence intervals in terms of pivotal quantities and quantile functions. AC&CS. 2021;55:66–84.
  16. Nechval NA, Berzins G, Nechval KN. A new technique of invariant statistical embedding and averaging in terms of pivots for improvement of statistical decisions under parametric uncertainty. In: Arabnia HR, editor. Advances in Parallel & Distributed Processing, and Applications, Transactions on Computational Science and Computational Intelligence; CSCE'20-The 2020 World Congress in Computer Science, Computer Engineering, & Applied Computing; 2020 July 27-30; Las Vegas, USA, Switzerland AG: Springer Nature; 2021. p. 257–274.
  17. Nechval NA, Berzins G, Nechval KN. A new pivot-based approach to constructing prediction limits and shortest-length or equal tails confidence intervals for future outcomes under parametric uncertainty. In: Bruno Castanier, Marko Cepin, David Bigaud, editors. Proceedings of the 31st European Safety and Reliability Conference (ESREL 2021); 2021 September 19-23; Angers, France: Research Publishing; 2021. p. 2886–2893.
  18. Nechval NA, Berzins G, Nechval KN, et al. Intelligent constructing efficient statistical decisions via pivot-based elimination of unknown (nuisance) parameters from underlying models. AC&CS. 2021;55:469–489.
  19. Nechval NA, Berzins G, Nechval KN, et al. Innovative technique for computing shortest length and/or equal tails confidence intervals in reliability and safety under parametric uncertainty. Pattern Recognition and Information Processing (PRIP'2021): Proceedings of the 15th International Conference; 2021 Sept 21–24; Minsk, Belarus: 2021. p. 206–210.
  20. Nechval NA, Nechval KN, Berzins G. A new unified computational method for finding confidence intervals of shortest length and/or equal tails under parametric uncertainty. Proceedings of the 2021 International Conference on Computational Science and Computational Intelligence (CSCI); 2021 December 15-17; Las Vegas, NV, USA: IEEE; 2021.p. 533 – 539.
  21. Nechval NA, Berzins G, Nechval KN. A new simple computational method of simultaneous constructing and comparing confidence intervals of shortest length and equal tails for making efficient decisions under parametric uncertainty. In: Yang XS, Sherratt S, Dey N, editors. Lecture Notes in Network and Systems (LNNS): Proceedings of Sixth International Congress on Information and Communication Technology–ICICT 2021; 2021 February 25-26; London, United Kingdom: Springer Nature; 2022. p. 473–482.
  22. Nechval NA, Berzins G, Nechval KN, et al. A new intelligent method for eliminating unknown (nuisance) parameters from underlying models as an alternative to the Bayesian approach. Journal of Applied Mathematics and Computation. 2022;1:53–65.
  23. Nechval NA, Berzins G, Nechval KN. Intelligent computational approach to constructing adequate statistical decisions under parametric uncertainty of applied stochastic models. Proceedings of the 2022 International Conference on Computational Science and Computational Intelligence; 2022 December 14-16; Las Vegas, United States: IEEE; 2023. p. 522–529.
  24. Nechval NA, Berzins G, Nechval KN. Adequate mathematical models of the cumulative distribution function of order statistics to construct accurate tolerance limits and confidence intervals of the shortest length or equal tails. WSEAS Transactions on Mathematics. 2023;20:154–166.
  25. Nechval NA, Berzins G, Nechval KN. Optimal statistical estimation and dynamic adaptive control of airline seat protection levels for several nested fare classes under parametric uncertainty of customer demand models. WSEAS Transactions on Mathematics. 2023;22:395–408.
  26. Nechval NA, Berzins G, Nechval KN. A unified technique for prediction and optimization of future outcomes under parametric uncertainty via pivotal quantities and ancillary statistics. AC&CS. 2023:57(3):234–257.
  27. Nechval NA, Berzins G, Nechval KN. A novel computational intelligence approach to making efficient decisions under parametric uncertainty of practical models and its applications to industry 4.0. In: Advanced Signal Processing for Industry 4.0. IOP Publishing Ltd; 2023;1(7):1–40.
  28. Nechval NA, Berzins G, Nechval KN. New applications of computational intelligence for constructing predictive or optimal statistical decisions under parametric uncertainty. AC&CS. 2023;57(5):473–489.
  29. Nechval NA, Berzins G, Nechval KN, et al. Adequate intelligent constructing predictive or optimal statistical decisions under parametric uncertainty of applied stochastic models. International Journal of Applied Mathematics and Machine Learning. 2023:17(1):43–81.
Creative Commons Attribution License

©2024 Nechval, et al. This is an open access article distributed under the terms of the, which permits unrestricted use, distribution, and build upon your work non-commercially.