Kendall did that even before Moran’s work (1954) appeared. Herestricted his attention to such values <strong>of</strong> <strong>the</strong> parameters a <strong>and</strong> b <strong>in</strong>formula (3.8) which determ<strong>in</strong>e a stationary process, <strong>and</strong> he mostlyworked with series from economics. Such series rarely oscillate aroundone level creat<strong>in</strong>g a stationary process. They usually have a tendency,a trend. The production <strong>of</strong> electrical energy, say, <strong>in</strong>creasesexponentially <strong>and</strong> <strong>the</strong>refore has a l<strong>in</strong>ear trend when described on alogarithmic scale. The problem consists <strong>in</strong> describ<strong>in</strong>g <strong>the</strong> deviationsdur<strong>in</strong>g different years from <strong>the</strong> general tendency.Kendall thought it possible to determ<strong>in</strong>e <strong>the</strong> trend by some method(but certa<strong>in</strong>ly not by naked eye which is too subjective for a rigorousstatistical school) <strong>and</strong> to subtract it. This additionally complicates <strong>the</strong>statistical structure <strong>of</strong> <strong>the</strong> rema<strong>in</strong><strong>in</strong>g deviations, but <strong>the</strong>re is noth<strong>in</strong>g tobe done about it. Exactly such deviations as though form<strong>in</strong>g astationary process were studied by <strong>the</strong> method <strong>of</strong> autoregression.It is difficult to pronounce a def<strong>in</strong>ite op<strong>in</strong>ion about his results. Insome cases <strong>the</strong> statistical tests were happily passed, but not <strong>in</strong> o<strong>the</strong>rcases. May we consider that success was really achieved <strong>in</strong> thoseformer or should we expla<strong>in</strong> it only by <strong>the</strong> small number <strong>of</strong>observations? And no explanation is known why, for example, <strong>the</strong>model <strong>of</strong> autoregression with <strong>the</strong> trend be<strong>in</strong>g elim<strong>in</strong>ated does not suit<strong>the</strong> series <strong>of</strong> <strong>the</strong> cost <strong>of</strong> wheat but suits <strong>the</strong> total head <strong>of</strong> sheep. Nodecisive success <strong>in</strong> treat<strong>in</strong>g economic series was thus achieved.Kendall (1946) <strong>in</strong>vestigated <strong>the</strong> process <strong>of</strong> autoregressionconstructed accord<strong>in</strong>g to equation (3.8) by means <strong>of</strong> tables <strong>of</strong> r<strong>and</strong>omnumbers; <strong>the</strong> longest <strong>of</strong> <strong>the</strong> modelled series had 480 terms. Inconclud<strong>in</strong>g, let us have a look at <strong>the</strong> empirical estimate <strong>of</strong> a correlationfunction (Fig. 3, dotted l<strong>in</strong>e). See how much <strong>the</strong> estimate differs from<strong>the</strong> real values (cont<strong>in</strong>uous l<strong>in</strong>e) <strong>and</strong> fades considerably slower than<strong>the</strong> real function.Hannan (1960) published an estimate <strong>of</strong> <strong>the</strong> spectral density <strong>of</strong>Kendall’s series. The graphs <strong>of</strong> <strong>the</strong> <strong>the</strong>oretical density <strong>and</strong> its variousestimates are shown on Fig. 4. It is seen that <strong>the</strong>y are pretty littlesimilar to <strong>the</strong> true density. In particular, <strong>the</strong> later takes a maximalvalue near po<strong>in</strong>t λ = π/5 whereas <strong>the</strong> maximal values <strong>of</strong> all <strong>the</strong>estimates are at po<strong>in</strong>t λ = π/15.An unaccustomed eye can imag<strong>in</strong>e that small values <strong>of</strong> <strong>the</strong> spectraldensity are estimated well enough, but noth<strong>in</strong>g <strong>of</strong> <strong>the</strong> sort is reallytak<strong>in</strong>g place. The relative error is here just as great as <strong>in</strong> <strong>the</strong> left side <strong>of</strong><strong>the</strong> graph, i. e., as for large values <strong>of</strong> <strong>the</strong> density. We see that <strong>the</strong>correlation <strong>the</strong>ory, created by <strong>the</strong> founders <strong>of</strong> <strong>the</strong> <strong>the</strong>ory <strong>of</strong> stochasticprocesses for treat<strong>in</strong>g discrete observational series, such as <strong>the</strong> number<strong>of</strong> sunspots <strong>in</strong> various years or <strong>the</strong> values <strong>of</strong> economic <strong>in</strong>dicatorsexactly <strong>in</strong> those cases did not atta<strong>in</strong> undoubted success.The idea <strong>of</strong> a ma<strong>the</strong>matical description <strong>of</strong> wavy processesencountered <strong>the</strong> practical difficulty <strong>in</strong> that any proper estimation <strong>of</strong> <strong>the</strong>correlation function dem<strong>and</strong>s not tens or hundreds <strong>of</strong> separateobservations, but (Kendall 1946) tens <strong>and</strong> hundreds <strong>of</strong> pert<strong>in</strong>ent waveswhich means thous<strong>and</strong>s <strong>and</strong> tens <strong>of</strong> thous<strong>and</strong>s observations. On <strong>the</strong>o<strong>the</strong>r h<strong>and</strong>, parametric models such as <strong>the</strong> model <strong>of</strong> autoregression hadnot been conv<strong>in</strong>c<strong>in</strong>gly statistically confirmed. Consequently, <strong>the</strong>74
applications <strong>of</strong> <strong>the</strong> <strong>the</strong>ory <strong>of</strong> stochastic processes to that material, <strong>and</strong>to forecast<strong>in</strong>g <strong>in</strong> particular, are not sufficiently scientifically justified.The worst circumstance is that many contributions are published <strong>in</strong>that field such as Ivakhnenko & Lapa (1971) which do not sufficientlycheck <strong>the</strong> adopted model statistically <strong>and</strong> <strong>the</strong>refore can not beconsidered seriously.The situation would have been quite bad but at <strong>the</strong> same time newfields <strong>of</strong> application <strong>of</strong> <strong>the</strong> correlation <strong>the</strong>ory <strong>in</strong> aero-hydrodynamics<strong>and</strong> physics which constitute <strong>the</strong> real worth <strong>of</strong> that <strong>the</strong>ory werecreated. We will <strong>in</strong>deed consider <strong>the</strong>se applications.3.5. Processes with stationary <strong>in</strong>crements. When hav<strong>in</strong>g somema<strong>the</strong>matical tool <strong>and</strong> wish<strong>in</strong>g to describe natural phenomena by itsmeans, <strong>the</strong> most important consideration is, not to ask nature for toomuch, not to attempt to apply that tool <strong>in</strong> cases <strong>in</strong> which it is helpless.Thus, when imag<strong>in</strong><strong>in</strong>g some wavy phenomenon, we would have likedto apply <strong>the</strong> <strong>the</strong>ory <strong>of</strong> stationary stochastic processes for describ<strong>in</strong>g it.However, it was gradually understood that <strong>the</strong> largest waves <strong>in</strong> <strong>the</strong>observed process can ei<strong>the</strong>r be not <strong>of</strong> a statistical essence at all, or thatour observations conta<strong>in</strong> <strong>in</strong>sufficient data for determ<strong>in</strong><strong>in</strong>g <strong>the</strong>irstatistical characteristics, or, f<strong>in</strong>ally, that a purely statistical descriptioncan be short <strong>of</strong> our aims.For example, <strong>the</strong> cyclic recurrence <strong>of</strong> economic life apparently hasall <strong>the</strong>se <strong>in</strong>dications. Here, we can not on pr<strong>in</strong>ciple consider aphenomenon as statistical because only one realization <strong>and</strong> nostatistical ensemble is available. And <strong>of</strong> course we usually have<strong>in</strong>sufficient observations. F<strong>in</strong>ally, a statistical description does notsatisfy us because we need to know, for example, not how one orano<strong>the</strong>r decl<strong>in</strong>e or rise is develop<strong>in</strong>g <strong>in</strong> <strong>the</strong> mean but what happenswith <strong>the</strong> particular decl<strong>in</strong>e or rise exist<strong>in</strong>g this moment.It is absolutely impossible to reckon on describ<strong>in</strong>g phenomena <strong>of</strong><strong>the</strong> largest scale <strong>in</strong> <strong>the</strong> boundaries <strong>of</strong> <strong>the</strong> <strong>the</strong>ory <strong>of</strong> stochastic processes.The situation is different for phenomena on a small scale; <strong>in</strong> such casesperhaps someth<strong>in</strong>g can be done. Take ano<strong>the</strong>r example, <strong>the</strong> course <strong>of</strong>meteorological processes. It is absolutely clear that a statisticaldescription <strong>of</strong> <strong>the</strong> largest changes <strong>of</strong> <strong>the</strong> wea<strong>the</strong>r on a secular scale isimpossible <strong>and</strong> senseless. It is uncerta<strong>in</strong> beforeh<strong>and</strong> whe<strong>the</strong>r statisticalmethods can be applied for describ<strong>in</strong>g changes <strong>of</strong> <strong>the</strong> wea<strong>the</strong>r on asmall scale dur<strong>in</strong>g a few days, for predict<strong>in</strong>g it, say. However,experience shows that this is sufficiently useless. Still, when restrict<strong>in</strong>gforecasts to small territories <strong>and</strong> short <strong>in</strong>tervals, <strong>the</strong> success <strong>of</strong>statistical methods is brilliant. The relevant <strong>the</strong>ory is called statisticalKolmogorov – Obukhov <strong>the</strong>ory <strong>of</strong> turbulence <strong>and</strong> we will later say afew words about it.We turn now to geology <strong>and</strong> formulate, for example, <strong>the</strong> problem <strong>of</strong>estimat<strong>in</strong>g <strong>the</strong> reserves <strong>of</strong> a deposit given <strong>the</strong> per cent <strong>of</strong> <strong>the</strong> usefulcomponent <strong>in</strong> a number <strong>of</strong> sample po<strong>in</strong>ts. Here also we encounter <strong>the</strong>risk <strong>of</strong> apply<strong>in</strong>g stationary processes for describ<strong>in</strong>g that per cent over<strong>the</strong> entire deposit. The situation with <strong>the</strong> ensemble <strong>of</strong> realizations <strong>and</strong><strong>the</strong> availability <strong>of</strong> data is very bad for determ<strong>in</strong><strong>in</strong>g <strong>the</strong> statistics <strong>of</strong> <strong>the</strong>largest fluctuations. On <strong>the</strong> o<strong>the</strong>r h<strong>and</strong>, <strong>the</strong> largest irregularities occuron a large scale <strong>and</strong> likely change smoothly; it may be <strong>the</strong>refore75
- Page 1 and 2:
Studies in the History of Statistic
- Page 3 and 4:
Introduction by CompilerI am presen
- Page 5 and 6:
(Lect. Notes Math., No. 1021, 1983,
- Page 7 and 8:
sufficiently securely that a carefu
- Page 9 and 10:
is energy?) from chapter 4 of Feynm
- Page 11 and 12:
demand to apply transfinite numbers
- Page 13 and 14:
for stating that Ω consists of ele
- Page 15 and 16:
chances to draw a more suitable apa
- Page 17 and 18:
Let the space of elementary events
- Page 19 and 20:
2.3. Independence. When desiring to
- Page 21 and 22:
Eξ = ∑ aipi.Our form of definiti
- Page 23 and 24: absolutely precisely if the pertine
- Page 25 and 26: where x is any real number. If dens
- Page 27 and 28: probability can be coupled with an
- Page 29 and 30: Nowadays we are sure that no indepe
- Page 31 and 32: λ = λ(T)with λ(T) being actually
- Page 33 and 34: (1/B n )(m − A n )instead of the
- Page 35 and 36: along with ξ. For example, if ξ i
- Page 37 and 38: µ( − p0) ÷np0 (1 − p0)nhas an
- Page 39 and 40: distribution of the maximal term |s
- Page 41 and 42: ξ (ω) + ... + ξ (ω)n1n{ω :|
- Page 43 and 44: P{max ξ(t) ≥ x} = 0.01, 0 ≤ t
- Page 45 and 46: 1. This example and considerations
- Page 47 and 48: IIV. N. TutubalinTreatment of Obser
- Page 49 and 50: structure of statistical methods, d
- Page 51 and 52: Suppose that we have adopted the pa
- Page 53 and 54: and the variances are inversely pro
- Page 55 and 56: It is interesting therefore to see
- Page 57 and 58: is applied with P(t) being a polyno
- Page 59 and 60: ut some mathematical tricks describ
- Page 61 and 62: It is clear therefore that no speci
- Page 63 and 64: of various groups of machines, and
- Page 65 and 66: nnA(λ) x sin λ t, B(λ) = x cosλ
- Page 67 and 68: of the mathematical model of the Br
- Page 69 and 70: dF(λ) = f (λ) dλ, so that B( t
- Page 71 and 72: usually very little of them. Indeed
- Page 73: This is the celebrated model of aut
- Page 77 and 78: achieved by differentiating because
- Page 79 and 80: u(x 1 , x 2 , t 1 , t 2 ) = v(x 1 ,
- Page 81 and 82: Reasoning based on common sense and
- Page 83 and 84: answering that question is extremel
- Page 85 and 86: IIIV. N. TutubalinThe Boundaries of
- Page 87 and 88: periodograms. It occurred that work
- Page 89 and 90: at point x = 1. However, preceding
- Page 91 and 92: He concludes that since the action
- Page 93 and 94: The verification of the truth of a
- Page 95 and 96: In the purely scientific sense this
- Page 97 and 98: ought to learn at once the simple t
- Page 99 and 100: the material world science had inde
- Page 101 and 102: values of (2.1) realized in the n e
- Page 103 and 104: *several dozen. The totality µ ica
- Page 105 and 106: Mendelian laws. It is not sufficien
- Page 107 and 108: example, the problem of the objecti
- Page 109 and 110: a linear function is not restricted
- Page 111 and 112: 258 - 82 - 176 cases or 68.5% of al
- Page 113 and 114: The Framingham investigation indeed
- Page 115 and 116: or, for discrete observations,IT(ω
- Page 117 and 118: What objections can be made? First,
- Page 119 and 120: eliability and queuing are known to
- Page 121 and 122: Kolman E. (1939 Russian), Perversio
- Page 123 and 124: measurement is provided. Recently,
- Page 125 and 126:
which means that sooner or later th
- Page 127 and 128:
The foundations of the Mises approa
- Page 129 and 130:
A rather subtle arsenal is develope
- Page 131 and 132:
4.3. General remarks on §§ 4.1 an
- Page 133 and 134:
BibliographyAlimov Yu. I. (1976, 19
- Page 135 and 136:
processes are now going on in the s
- Page 137 and 138:
obtaining a deviation from the theo
- Page 139 and 140:
VIOscar SheyninOn the Bernoulli Law