Classification – We learned what was supervised learning and we know that classification is one branch of supervised learning. So we have this definition of classification like it is the task of learning Target function. This is also the classification model that map’s each of the attribute set that is X to one of the predefined class labels y. Like we have some attribute set say for example where you have x 1 to xn and you just put all of this for classification model this how we build that we discussed in our previous article.
Now, this classification model will be classified into different class labels. Is it already has so consider say for example, we have a dataset. So this is our data set which consists of a number of rows and number of columns. Now we to be bifurcation or be divided by 67 per cent or thirty-three per cent or it can be like 50 per cent or 50 per cent now this will be our training site. All right. This is going to be a test set.
So it’s training set what we do is we construct a model. So that model is basically a classification model and to this we put an input attribute set. So that is represented by X here and for this to be had some class labels. So that is here. Y. So what is the already we have seen for example this data set. So this data set basically consists of data objects. Now you might be wondering what this data object is. So that is nothing but it can be vector or we can be some cases or it can be some observations from a sample set or it can be some entity It can be anything like this.
So now this each of these data objects has got certain kind of attibute. So it can be like in case of the weight mass is an attribte or in case of year time can be attribute or it can be anything so attribute have got different properties. And so it is of two types. That is he have Discrete and continuous So it can be either qualitative and quantitative attributes. So those attributes you just input into this classification model and for that reproduces in class levels, which are already there.
Now, this classification basically has got one classifier. So what is classifier is a model that create from here and this model reflects what makes the instances in one class different from the other class say we have two classes class 1 and Class 2. We have a positive class and a negative class. Plus we have some data objects assigned here. We have some data objects assigned here and say five and five. So this classifier basically does is bifurcation or basically puts these data objects into these many classes so our classifier mainly would be decision trees.
So we can be like default class and from that, you have some branches, and so will be in a number of others and so you traverse this decision tree to create some rules. Now you have naive Bayes classifier. So that it’s mainly dealt with the probability distribution so you have some probability of an event based on some condition. And so you calculate the condition, given that probability and is multiplied by the event divided by the condition. So, NB and decision trees are the most commonly used classifiers in classification and the other classifiers which can be used as you have KNN i.e K-nearest neighbours.
Or you can have the support vector machines (SVM). So these are the basic, classifiers which are used and so classified is mainly denoted with the help of a symbol called eta. So this classifier is mainly used in order to classify different objects into different class labels. Now, there are two major phases of classification that is we have the learning phase and we have the querying phase So what learning phase is also called as the training phase.
So these basically set of mutually exclusive classes now, we can be like c1…ck like what class1, class2, class3 up till classN. Then it can be assigned to a population through this alphabet for this calligraphic symbol represents a population which you need to study. So this is a population under study. So we have all these classes assigned to this label that is this population. So each of these classes have each of the instances that is xi…xn we assign it to a training set C for the population under study. Now, this is the learning phase where the classifier will learn what kind of classes it has.
And what are the various input attribute set it has got. Now, it will do the querying phase so that it step 2. So first you have learning phase then you have any querying phase so what querying phase what the classifier does is it takes each of the input attribute set X which belongs to this training set D. i.e. the population under study and it will check whether that classifier or whether the input attribute set belongs to this which of this classes from the population under study means if there is one attribute say, for example, you have attribute 1 and it is from this population under study.
It will check whether this X or this attribute belongs to which classes, say, for example, if x belongs to class even then it will be labelled as C1. Let’s understand with the help of an example now say, for example, you have a credit card value limit. So for example, your particular customers and you have the credit limit say for example $100 to $10K. Now, what classification it does is.
It will assign each of this to a set like say for example $100-$200 assigns to class1 Second. $200-$300 It assigns to class2. Say For example $300 to $400 class3 likewise till Class N. So what basically is we have a large number of data values which was quantitative values and we just bifurcated into small sets. And this would you do different classes. So regression would be the opposite of this we just open this so that we can have a chain of different value in this So well that was all regarding classification in ML.
reference – Classification
Web enthusiast. Thinker. Evil coffeeaholic. Food specialist. Reader. Twitter fanatic. Music maven. AI and Machine Learning!