#prior probability

[nofatclips]

3blue1brown on probability vs odds and the medical test paradox

[ijtsrd]

Mining Educational Data to Predict Students' Future Performance using Naïve Bayesian Algorithm

by Nilaraye ""Mining Educational Data to Predict Students' Future Performance using Naïve Bayesian Algorithm""

Published in International Journal of Trend in Scientific Research and Development (ijtsrd), ISSN: 2456-6470, Volume-3 | Issue-5 , August 2019,

URL: https://www.ijtsrd.com/papers/ijtsrd26642.pdf  

Paper URL: https://www.ijtsrd.com/engineering/computer-engineering/26642/mining-educational-data-to-predict-students%E2%80%99-future-performance-using-na%C3%AFve-bayesian-algorithm/nilaraye

ugc approved journals for social science, management journal, paper publication for engineering

Higher education institutions want not only to provide quality education to its students but also to advice career options according to the prediction of students' performance. The students' satisfactory performance takes an important role to give birth the best quality graduates who will become competent laborers for the country's economic and social development 2 . Students' performance like who will pass and who are likely to fail can be predicted with the help of lots of features available. The students want to realize their final performance before the announcement of their results and before they attend their semester exams. According to their predicted performance, the students can improve their skills by proper planning to lead to a good performance in their end examination. To provide a good advice to such kind of student, educational data mining system is implemented to predict students' final performance evaluated by considering factors which include IM, PSM, Basics, ACIC, ASS, CP, ATT, ACOC and ESM. In this research, an attempt has been made to explore Naïve Bayesian classification to predict the students' future performance. 

[mavaddat]

[L]ikelihoods don’t tell you what to believe or how to act or which hypotheses are probably true; they merely tell you how to compare the degree to which the evidence supports the various hypotheses you wish to consider (Edwards 1972, Royall 1997). Likelihoodism is sometimes criticized for entailing that perfectly absurd hypotheses often have likelihoods that cannot be bettered. [By likelihoodism, I mean the comparative principle that O supports H1 more than O supports H2 if and only if Pr(O * H1) > Pr(O * H2). It is a further claim that degree of differential support is measured by the likelihood ratio. Formulations of the Likelihood Principle often combine these two elements; see Forster and Sober (2002) for discussion.] If you draw the six of spades from a deck of cards, the hypothesis that this was due to the intervention of an evil demon bent on having you draw that very card has a likelihood of unity [i.e., likelihood ≐ 1], but few of us would regard this hypothesis as very plausible. Doesn’t it sound strange to say that your drawing the six of spades supports the demon hypothesis more than it supports the hypothesis that the card was drawn at random from a normal deck? Yet this is precisely what likelihoodism asserts. Whatever the merits of this objection, it is not something that a Bayesian should embrace. The reason is that Bayes’s theorem tells us that the observation of the six of spades confirms the demon hypothesis, in the sense that it raises its probability. This is the familiar point that when a hypothesis entails an observation, and the observational outcome was not certain to occur, and the hypothesis’s prior probability is neither zero nor one, the observation confirms. [Since Bayesians usually reserve priors of 0 and 1 for tautologies and contradictions, I take it that they will want to assign the demon hypothesis an intermediate prior probability.] It is entirely consistent with this point that the probability of the demon hypothesis remains very low and the normal hypothesis’ probability remains very high. But if confirmation concerns the diachronic question of how probabilities change rather than the synchronic question of what a probability’s absolute value is, then Bayesians have to concede that the observation of the six of spades confirms the demon hypothesis. If so, they should not cast a jaundiced eye on the likelihoodist’s claim about differential support. Likelihoodists can and should admit that the demon hypothesis is implausible or absurd, notwithstanding the fact that it has a likelihood of unity [i.e., likelihood ≐ 1] relative to the single observation under consideration. It’s just that likelihoodists decline to represent this epistemic judgment by assigning the hypothesis a probability. Likelihoodist epistemology is modest in its ambitions; support gets represented formally, but plausibility does not. It thereby contrasts with (strong) Bayesianism, which, as I’ve explained, aims to characterize all genuine epistemic concepts.

"Bayesianism -- Its Scope and Limits" in R. Swinburne, ed., Bayes' Theorem, Proceedings of the British Academy Press, vol. 113, 2002, pp. 21-38.