Okay this is naive Bayes classification algorithm. So, for Naive Bayes classification algorithm is a is a classification technique is using these Bayes theorem. So the variables must be independent from each other Naive Bayes model is easy to do. And Bayes theorem is something like a probability of cross given x equal to probability of X given cross times the probability of cross divided by the probability of x. So, probability of cross given x probably to have cross probability of X given cross probability of x. So, for Naive Bayes classification example, let's say we have a data set of 1500 records with tree crosses.
Para giant fish. So it cross we have around 500 rows. So all the variables are categorical variables we either either true or false. So we create a frequency table from our data. So for the parent we have around 500 rows Don't we have 500 rows and fish we have 500 rows or so. So for let's say for stream we have around 54.
We have our 500 or 500 rows. For green we have 405 ellos. For dangerous t we have zero final rules for para cross for W for up to 500 rules that can stream for wins we have zero green color we have zero by 100 those dangerous to via via via rough or cross for fish. We have 500 firewall can stream we observe our firewall rules. Green color, we have 100 our firewall rules. Dangerous t we have 5500 rules for the fish class.
So, we want to predict data. The same is true Da Vinci's false color is true, the dangerous T is false. So, we want to predict what class y Nima Tyrese. So we can use this Naive Bayes algorithm. So, this is the same RNA based algorithm with a previous one. I will say this Naive Bayes algorithm is much more easier to understand.
So, probability of the hypothesis given multiple evidences is equal to the probability of evidence one, even the hypothesis time is the probability of an evidence to, given the hypothesis times until the property probability of evidence and given a hypothesis times probability of the hypotheses, divided by the probability of multiple evidences. So for our, for our data, we will calculate a probability of hypotheses will calculate probability of hypotheses. That crosses.so probability of dot given streaming green is equal to probability of stream given is a dot times probability of green given is a dot times probability of da divided by probability of stream in green. So we have around zero for hypothesis testing on the cross to be Para USB probability or para given streaming green, equal to probability of seeing, given para times the probability of green given para, times the probability of para divided by probability of streaming green.
So it will be around 0.0264 divided by probability of swimming green. For the hypothesis testing for cluster B fish will be probability of fish given swimming, green equal probability, probability of swim, given a fish times probability of green given is a fish times probability of fish divided by probability of the streaming green. And we have around 0.06 they say divided by the probability of swimming green. So the denominator is probability of swimming green is the same in all calculations. and predict across all will be fish because fish has a higher probability. So fish we have around 0.06, a six, then a parody of 0.026.
For dog we have zero so fish has the highest probability. So we predict the data to be fish. So this is how the naive Bayes algorithm works.