I stand at the fork of decisions, and think how far I can see beyond the undergrowth. Sadly, my eyes fail
before the expanse of the horizon or the undergrowths veil my mind. But, wait, there comes my friend –
Thomas Bayes – and I suppose he can handle this with his strange play of numbers.
Welcome to Bayesian Statistics, dear friends!
When less is known of the matter in question, but more is required of the answer to supply, Bayes solves
where Frequentist methods fail to apply. Too much of jargon, eh! Fine, I will simplify that for you.
When we talk of a dependent variable’s variance in the data of the future, but we have less data of the
past to exactly predict it, or less of independent variables to analyze the situation, we use Bayesian
Statistics. It uses the statistical history of the independent variables and individually acts on them to
predict the further changes in the dependent variable. For example, I need to do some background check
on a person, but I don’t know much of his past or a large number of friends. So, in this case, I go for a
check of his friends’ background checks from the day of his being in relation with them, and by this I
make a probable forecast of how the person would be in the days to come.
Coming to the technical side, Bayesian statistics involves an axiomatic approach to a statistical question.
It is an interpretation of probability as a ‘rational, conditional measure of uncertainty’. With objective
methods available to analyze several situations, it is possible to handle seemingly incoherent scientific
hypotheses in an assembled manner of analysis using Bayesian Statistics, while this possibility cannot
even be thought of in Frequentist Statistics.
Sometimes, Bayesian Statistics methods are referred to be ‘probabilities of probabilities’ corresponding to
the mathematics style followed in its usage. Further, the mathematics involved in it is dynamic in nature,
and adjusts itself to the changes in the data-set. Bayesian methods in a way reduce statistical inference to
problems in probability theory, and serve to prove the point stated either by logical justification or by
proving the logical inconsistency of others.
Apart from the mathematics involved in Bayesian Statistics, a set of scholars work on the non-calculative
section based on the ‘degrees of belief’. This section analyzes an event in a manner of how many people
talk of it, how they talk of it (qualitative), what is the context that they talk of, etc. An equivocal situation
arises in the accuracy of this method, and application of this in real-time situations is hence restricted.
With the prior information and limited indirect knowledge of the variable, Bayesian Statistics helps in the
analysis. It is dynamic in the prediction of the results, and works in accordance to the trends of the
independent variables. This ability of Bayesian Statistics to adapt to the changes in the situation makes it
more resourceful.
Coming to the field of applications, we shall be surrounded by many, but what about money rather than
many. Yes, my friends, you read it right. Bayesian Statistics is the greatest concept of the money-printing
machine known as the Stock Market. Goldman Sachs, Franklin Templeton, AOS, you name it and they
use it. The use of Bayesian Statistics outnumbers the rate of success of any other method to predict the
market. So fluid, so dynamic, so uncertainly certain it is – that mind trusts it for money.
In reality, we use this form of prediction every moment in life, weighing one situation to another, and
thinking of others’ remarks and how they would have done. So, with a crude form of Bayesian Statistics
running in our minds and helping us understanding things around us, we walk unknown yet well-known
to this mathematical concept.