Let’s start with the session of this model this algorithm is based on the base theorem it is used for classification again it uses probability values to decide which class a test point belongs to this is the only algorithm that gets this term associated with it called naive the term 9 is not a very good thing to be associated with ok, but this algorithm is the only one which gets associated with that very unfortunate because all algorithms are native.
As per this definition, the reason why it’s called naive is this algorithm assumes that all the variables that you’re going to use to build the models of those variables are independent of one another the independence assumption is often violated we already discussed this but then this is true for all the other ones it’s called naive due to the assumption that the features in the data set are mutually independent they don’t interact with each other, but that’s what every algorithm does.
This is the only one which gets the cap the hat whatever they provide but the beauty of this algorithm is what this algorithm says is howsoever weak predictors your features maybe don’t drop them consider them take into account the information given each feature and build your model based on probabilities so it doesn’t advocate you from dropping any of the features how they were significant the overlap might be using them that was this algorithm to understand this algorithm we have to understand what is the probability we have to understand what is joint probability.
We have to understand what is conditional probably three of them so let’s quickly start what is a probability how do you define probabilities okay chances of some event occurring so in the definition you have an event and then you have something called chances what is these chances how do you find that out how do you care to find the chances of something happening so what involves the Experian what is there in the experimental trials and proportion right so if I ask you what is the probability that today it will rain and you might be surprised just a week ago a week or 10 days ago there was news on the web wherein Kerala there were snowfalls I’ve seen that in Kerala Kerala there’s a hill hilly a body called Hill spot moon are right there they had snowfall and the temperature was minus 3 right, so such events can happen.
If I ask you what is the probability that today it’s going to rain at 5 o’clock when it is easy who said you know to take the weather just talk so what you’ve done is you have taken your experience into account on this particular date January is Jan month of January almost middle of January how many times in the past did I see rain that ratio how many times you saw this event of interest versus your records the number of trials this ratio is called frequencies approach to probability.
The probability is a ratio well it’s a ratio of the events of interest to you how many times they have occurred out of the total number of trials how many times they even could have occurred that is the ratio called probability values did you know that probability values can be calculated in another way have you done Poisson distribution in statistics Poisson distribution very convoluted formula there if given the state I am in what is the probability that I’ll be in a different state the next state that is a function of how many ways you can reach that state, okay, so that is another way of finding probability we are going to use the previous def of probability which is based on frequencies ratios okay now that you know.
What is the probability the next thing you need to know is what is joint probability joint probabilities’ probability of multiple events occurring together okay so what is the probability of if I give you a deck of cards and asked you to pull a card out of it and I ask you what is the probability that this card is going to be a red king, so there are two colors red in black 52 cards so 52 cards you have two red King’s hearts and diamonds so the probability of this joint event of the red king is red is one king is one of the probability is two by 52 all right now when you’re pulling out the card somehow you come to know that it is a red card some of you come through it’s a red card now what is the probability this is a red king now you know it’s a red card.
Your scope is now limited to only red cards to by 26 is 1 by 13 so you have taken into account and evidence and information that you captured that it is a red card the moment you get that information you recalibrate your probability of red king that concept of recalibrating your probabilities based on the information that you’re gathering that concept is called Bayes theorem so what Mr. BAE’s insiders start with some probability the default probability values but to keep on recalibrating those probabilities the moment more and more information comes to you however the information we may be howsoever weak the information maybe don’t ignore any information and recolored calibrate your probability that is the philosophy or reasoning behind this Basin basing Mode’s are you okay.
So for this to understand this we need to know probability joint probability and what are conditional probabilities probe what is the probability of King given it’s red we know it’s already red card so now if only the event left to happen is king so what is the probability that this key card will be a king given it’s read that concept which we call two by 26 that concept is called conditional probability.
We started by joint probability what is the probability is a red king because we didn’t have any information it was 2 by 52 the moment we came to know it’s red then we switched to the conditional probability what are the probabilities King given its red right conditional probability is invalid if the joint probability is 0 if the probability of two events occurring together with a 0 we don’t talk about conditional probability what is the probability of drawing a green King there is no green King possibility so don’t even talk about conditional probabilities so we talk about conditional probability only when we have a joint probability greater than 0 all right.
Let’s see how all this was put in to build a model we already discuss what is probably what is joint probability and what is the conditional probability we already discuss this so I’m going to jump these slides this is the model I am sure you have seen this formless looks very familiar oh man you should put it on some kind of board and put it on look so likely we’re going to launch some rocket it’s a pretty easy, ok so what I’m going to do is I’m going to jump all this right back.
I’m going to play with animations this blue box represents all the flight information have captured from my experience all the hundred present flights which have taken I am representing at Bay blue box the dimensions of the box are inconsequential they don’t matter when I look at the past data about the flights which I have taken I noticed that 20% of the time those flight delay 80% of the time the flight was on the shadow in mathematics in school order we would have done this as Venn diagrams ok.