 In this video we will see three different papers where Descentry and Naibes have been used. Since I was explaining the basics of Descentry and Naibes using the simple example, simple data, let us look at the papers which use Descentry and Naibes in research. So, let us look at Naibes first. So, these papers, Descentry and Naibes has been used in research for long and there are a lot and a lot of publications on Descentry is available. You can check in the net in Google Scholar, but let us look at a very simple paper which is applicable here. What is the data they considered to predict the performance? So, to predict the performance they considered gender, their birth, it is for age, specialization, it is CTR from secondary school name status for this job, student status as a pass or incomplete student has, that is the thing we have to predict student status now. And the student is married or single, there is a marriage status and all the information is captured here. So, we have 1, 2, 3, 4, 5, 6, 7 features and one dependent variable to predict. So, with the cell values and they computed the frequency table, like a machine computes a frequency table, then they applied a Caneus neighbor and Descentry. Now, in Descentry, it is accurate, it is good, performance is good, that is the example. So, it is very simple. In a student's profile information is used and not their interaction system, not their attendance or not the marks in itself something, profile information is used to predict the students will pass the exam or not. So, very simple paper, you can collect this information from survey questions and you can use that. So, you can check this paper, how they did it, they explained Descentry in a detail that I gave this paper. So, just like just check this paper. Let us look at the next paper that is for Descentry. In this paper, the used Descentry to predict the student's performance. And so, the data they use is the nationality, student gender, the first language. So, what are the variables, decision variables, category variables in Descentry is also see the gender is true and nationality can be 2 or 4 and the first language is 1, 4 and teaching language, the high school percentage is like again 7. So, they have a multiple decisions to make and there are like 1, 2, 3, 4, 5, 6, 7, 8 different features, more than that. So, previous semester marks, weeks, friends and the father's occupation, mother's occupation, qualification of the father, student discounts, any parents working in the university, all this information has been used and they use this data to come up with a final dataset. This is a final dataset with all the decisions and also they given the values in each dataset like how they distributed across these particular values has been given. So, given this dataset, they want to predict student's performance. So, this is just a distribution of the values, they want to show how the data is distributed, this is descriptor index. And given the table, they want to predict student's mark in the current semester, they used to see 4.5 decision to algorithm with a 10 fold cross validation on datasets, the confusion matrix is shown here. Now you know what is cross validation, what is confusion matrix, because now you are using, we have seen what is confusion matrix for 2 or 3 variables. Now you are able to see what is confusion matrix and you are able to make sense out of it. Excellent to excellent classification is good, 46% like question but recall is less because there are a lot of excellent has been classified as very good. So, look at that and in a C4.5 classical algorithm, what is the criteria to split the tree is used is information gain ratio. And maximum size of trees should be 4, not more than that, you can say that minimal leaf size should be 1. And the depth is how far the tree can grow. So, they do not want to grow tree or something, they want to make the tree pruning, the confidence gives the pruning criteria also. So, they used C4.5 also ID3, it is exactly same information gain, maximum size, minimal gain, also depth also can be given and that value is this. And cart is classification and regression tree, as I mentioned, the decision tree also can be used as regression, that is what they did. So, they applied decision tree, different decision algorithms and they reported results here. So, this paper will help you to understand what are the features they collected, how they combine the features, how they made the decisions, how they use it to make a decision and what are the results to compare. So, this is one paper you can read and check it out. So, let us look at this paper. In this paper, it has both the decision tree and NaE base. So, in this paper, it is the demographic information of the participants like a female, male, age, graduate students and everything. And how many students registered, how many students dropped out. So, let us see. Yep. So, this, do you remember, this is the equation we saw in the NaE base classifier. And this also explains how NaE base assumption is helping you to make this base theorem. And let us look at it. Yeah. So, I want to show this one thing. So, they used the decision tree, NaE base, nearest to neighbor, NaE base, nearest to neighbor, decision tree and some neural network with relays. Do not worry about that. So, NaE base is performed like this. Hope you understand this corner. This is ROC curve. So, NaE base is not good. Also, the nearest neighbor is not the female. But decision tree performed well. But although the nearest neighbor, like a neural network might have performed well. So, I just want you to check these papers, understand what is the paper data you used and what are the classifiers they use, how they report the results. Hope now you are able to understand the paper, which will be detailed because now you know what is algorithm, how does the matrix. So, that helps you to understand the paper and it gives you identify the gap in the existing literature. And that might motivate you to collect data on your own and you can write your own paper, do some own research. So, that three papers are given in this in this slide, you can download and you can check those paper data from scholar.google.com. It is all available for you. If not, put that in the forum. We can give a link where you can download that. So, there are three papers. So, can you list down application of decision tree and nearest and also what data is required, what are the features required and what are the categories in the features, like what are the decision-making values in the features. This is down, this is based on you know what is decision tree, now what is nearest, now you have seen three applications of papers. Can you list down of the applications of decision tree in learning analytics or the data from the learning environment. So, there is no answer. It is just if you listed down, just go ahead and try, try collecting data if possible and apply these algorithms. If the results are good, different and different context or different than the existing research, please publish it in a good venue, international conference or journals. So, in this week we saw what is decision tree, what is NEI base. You understood what is NEI base and I talked about state transition sometime back in the diagnostic analytics. If you can combine, if you can remember both transition and base theorem, I would request you go and check Eden Markov model. It is kind of, if you know both, it is very easy to understand and very intuitive to go and next step will be that. But that is not part of this course. That is completely optional. So, I hope you understood what is decision tree and NEI base. As usual, if you do not understand decision tree and NEI base, go and check the videos available in the net. The idea here is to you to understand what is the concept or logic intuition begin these two classifiers. Not to understand all the mathematical or training parameters to be used in the classifiers. Thank you.