I am trying to evaluate a formula for a pattern recognition problem but am having difficulty understanding what an a-priori probability is.
Suppose I have a dataset with 4 classes. 25% of the examples are in class 1, 25% are in class 2, 40 % are in class 3, and 10 % in class 4.
I now need to calculate the a-priori probability for each class. Most of the definitions I have found basically say something like "A priori probability is calculated by logically examining a circumstance or existing information regarding a situation." and I have seen many examples using a coin toss where the apriori probability for heads and tails is .5 for each case.
With the coin toss example in mind, I am tempted to say the a-priori probability of each class is .25 since I have 4 classes.
Or do I take into account the dataset when determining this probability (e.g a priori probability for class 3 is .4)? To me that seems more like an empirical probability since it is based on data. I would greatly appreciate some clarification.