probability estimate

Terms from Artificial Intelligence: humans at the heart of algorithms

Page numbers are for draft copy at present; they will be replaced with correct numbers when final book is formatted. Chapter numbers are correct and will not change now.

We may not always know the precise probability of en event occuring and instead need to estimate this. For dexample, if we are not sure whether a coin is faor, we might toss it many times and then use the proportion of times it lands heads as an probability estimate. Note that this estimate is itself usually a stochastic value as it is based on sampling. Often the frequncy estimate often has a binomial distribution, so that if the true probability is p and we use N tries to estimate it, then the proportion we measure has mean p and variance p(1-p)/N. However, this only applies when the estimate is based on lots of independent trials. In some circimstances more complex ways are needed to produce a probability estiamte.

Used in Chap. 19: page 329