دانلود کتاب Philosophy Of Probability And Statistical Modelling – فلسفه احتمالات و مدل سازی آماری

توضیحات

Humans have been thinking in probabilistic terms since antiquity. They have

been thinking systematically and philosophizing about probability since the

seventeenth century. And they have been formalizing probability since the end

of the nineteenth century. The twentieth century saw intense philosophical work

done on interpreting probability, in a sort of attempt to find out its essence. The

twenty-first century, I argue, will bring a focus on more practical endeavours,

concerning mainly the methodologies of data analysis and statistical modelling.

The essence of probability, it turns out, lies in the diversity of its uses. So, the

methodological study of the use of probability is what brings humans closer to

a comprehensive understanding of its nature.

These and other ideas expounded in this Element developed out of a Marie

Curie project on probability and propensities that I carried out at the Institute of

Philosophy of the School of Advanced Study at London University during

201315. I came out of that project with the distinct impression that the study

of practice was of primary importance; and that much philosophy of probability

is still to come to terms with it. This Element is my first attempt at the bare bones

of a new research programme into the methodology of statistical modelling.

Most of the Element is devoted to justifying this methodology on the grounds

of practical involvement with the scientific modelling practice but also, I argue,

on account of the limitations of the traditional interpretative approaches to the

topic.

Thus, the first half of the Element (Sections 17) is entirely a state-of-the-art

review of the historiography of probability and its ensuing impact upon the

interpretative endeavour. This is fitting for a Cambridge Elements volume,

which allows for a profuse setting of the stage. And it is anyway needed in

order to understand why nothing other than a study of the practice of statistical

model building will do for a full understanding of objective probability. I first

explore (in Section 1) the dual character of the notion of probability from its

inception the subjective and objective aspects of probability that are essential

to any understanding the concept. The twentieth century brought in several

interpretations of probability. But one way or another, they all aim to reduce

probability to either subjective or objective elements, thus doing away with the

duality; and one way or another they all fail, precisely because they do away

with the duality. In the remaining sections in this half of the Element, I analyse

in detail the many objections against both the main subjective interpretations

(the logical and personalist or Bayesian interpretations), and the main objective

interpretations (the frequency and propensity interpretations). To make most of

these interpretations work, and overcome the objections, demands some

Downloaded from https://www.cambridge.org/core. IP address: 181.171.20.59, on 25 Dec 2020 at 13:42:48, subject to the Cambridge Core

terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/97811089858262

Philosophy of Probability and Statistical Modelling

acknowledgment of the complex duality of probability. This is by now widely

accepted, and the Element first reviews the roots and consequences of pluralism

about objective probability.

The second half of the Element (Sections 813) then centres upon the

objective aspects of probability, but now without any pretence of a reduction

of the whole concept. The discussion is focused entirely on objective probabil-

ity, and it contains most of the original material. I advance a number of novel

theses, which I defend in various original ways as well as proposing a number of

new avenues for research. The starting point is pluralist, and it accepts the

duality insofar as it argues that there are important matters of judgement in the

selection of crucial aspects of the application of objective probability in prac-

tice. Here, the critical distinction, advanced in Sections 8 and 9, is between the

traditional project to merely interpret probability and a distinct project to study

the application of probability. On the other hand, I go considerably beyond the

pluralism defended in the first half of the Element and, in Section 10, I embrace

novel forms of pluralism and pragmatism regarding objective probability.

The central idea of the second half, which also informs the Element as

a whole and looms large through most of its discussions, is what I have

elsewhere called the tripartite conception of objective probability (Surez,

2017a). This is the idea that the failure to reduce chance to either propensity or

frequency ought to lead to the acceptance of all three concepts as distinct,

insufficient yet necessary, parts of the larger notion of objective probability.

This tripartite conception is introduced in Section 10, which also assesses the

role of judgement and various subjective components. Sections 11, 12, and 13

are then devoted to modelling methodology, and the application of the tripartite

conception in statistical modelling practice in particular, in what I call the

complex nexus of chance (CNC). The thought running through these sections

is new and radical: objective probability is constituted by a thick array of

interlinked practices in its application; these are practices that essentially

involve the three distinct notions pointed to above; and since none of these

notions is theoretically reducible to any combination or set of the other two, this

means that the overall methodology remains unavoidably complex. There is

no philosophical theory that may explicate fully the concept of objective

probability, or chance, by reducing this complexity, and this already sheds

light on the limitations of the interpretations reviewed in the Elements first half.

Whats more, the second half of the Element also continues to illustrate the

fundamental duality of probability unearthed in the historiographical material

reviewed in the first seven sections. It does so in three different yet interrelated

ways. First of all, it leaves open that subjective elements may come into the

nature of the single-case chances that make up the tripartite conception.

Downloaded from https://www.cambridge.org/core. IP address: 181.171.20.59, on 25 Dec 2020 at 13:42:48, subject to the Cambridge Core

terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781108985826Philosophy of Science

3

Secondly, confirmation theory comes into the assessment of evidence for and

against different models. And, finally, there are irreducible subjective judge-

ments involved in the pragmatist methodology advocated in the later sections.

For instance, in Section 11 I argue that choosing the appropriate parametrization

of the phenomenon to be modelled is a critical part; and there is no algorithm or

automatic procedure to do this the choice of free parameters is subject to some

fundamentally subjective estimate of what is most appropriate in the context

for the purposes of the model at hand. Once again, the subjective and the

objective aspects of probability meet in fundamental ways (see Gelman and

Hennig (2017) as well as my response Surez (2017b) for an account of such

a merge in practice). Another related sense of subjectivism in statistical model-

ling is sometimes referred to as the art of statistical modelling and concerns

the choice of a correlative outcome or attribute space. There is nothing arbitrary

about this subjectivity though, since it answers precisely to specific pragmatic

constraints: it is a highly contextual and purpose-driven judgement.

On my view, each of the parametrizations of a phenomenon involves

a description of its propensities, dispositions, or causal powers. What is relevant

about propensities is that they do not fall in the domain of the chance functions

that they generate (Surez, 2018). Rather a propensity is related to a chance

function in the way that possibilities are related to probabilities: the propensity

sets the range of possible outcomes, the full description of the outcome space,

while the chance function defined over this space then determines the precise

single-case chance ascribed to each of these outcomes. A different paramet-

rization would involve a different description of the systems propensities,

perhaps at a different level of generality or abstraction (and no parametrization

is infinitely precise); and focusing on a different set of propensities may well

issue in a different set of possible outcomes, hence a different outcome space,

over which a different chance function shall lay out its probabilities. Since the

parametrizations obey pragmatic constraints that require appropriate judgements

within the context of application, it follows that the outcome spaces will corres-

pondingly depend on such judgements. In other words, a chance function is not

just a description of objective probabilities for objectively possible outcomes; it is

one amongst many such descriptions for a particular system, made relevant by

appropriate judgements of salience, always within a particular context of inquiry.

Here, again, the subjective and the objective aspects of probability merge.

————————————————————–

ترجمه ماشینی :

انسان ها از دوران باستان به صورت احتمالی فکر می کردند. آنها از قرن هفدهم به طور سیستماتیک درباره احتمال فکر می کردند و فلسفه می کردند. و آنها از اواخر قرن نوزدهم احتمال را رسمی می کنند. قرن بیستم شاهد کار فلسفی شدیدی بود که بر روی تفسیر احتمال انجام شد، به نوعی تلاش برای کشف ماهیت آن. من استدلال می‌کنم که قرن بیست و یکم بر تلاش‌های عملی‌تر تمرکز خواهد کرد، که عمدتاً مربوط به روش‌شناسی تجزیه و تحلیل داده‌ها و مدل‌سازی آماری است. به نظر می رسد که جوهر احتمال در تنوع کاربردهای آن نهفته است. بنابراین، مطالعه روش شناختی استفاده از احتمال، چیزی است که انسان را به درک جامع ماهیت آن نزدیک می کند. اینها و ایده‌های دیگری که در این عنصر بیان می‌شوند، از پروژه ماری کوری در مورد احتمالات و گرایش

---

 

tag : دانلود کتاب فلسفه احتمالات و مدل سازی آماری , Download فلسفه احتمالات و مدل سازی آماری , دانلود فلسفه احتمالات و مدل سازی آماری , Download Philosophy Of Probability And Statistical Modelling Book , فلسفه احتمالات و مدل سازی آماری دانلود , buy فلسفه احتمالات و مدل سازی آماری , خرید کتاب فلسفه احتمالات و مدل سازی آماری , دانلود کتاب Philosophy Of Probability And Statistical Modelling , کتاب Philosophy Of Probability And Statistical Modelling , دانلود Philosophy Of Probability And Statistical Modelling , خرید Philosophy Of Probability And Statistical Modelling , خرید کتاب Philosophy Of Probability And Statistical Modelling ,