 Welcome back to our learning antics tools course. In this week, we will talk about introduction to machine learning. So, first let us start what is machine learning. The aim of this course is not to teach machine learning in LA. But in order to understand the tools, algorithms we use in this course, you should know what is machine learning, what is the basics, what is supervised algorithm, what is the algorithms or learning methodologies. So, let us start with why we have to learn machine learning. So, you can take a time and think of machine learning why you have to learn. So, whatever device you are using a mobile tablet or a browser in your laptop or taking this course, machine learning algorithms are used to reduce the manual work. So, it is everywhere. So, we have to know what is machine learning and it is everywhere and it is developed to reduce human work. Imagine you are creating a system which classifies the road signals human has to go and classify the road signals for the automated car. If machine learning what it happens it can actually classify the road signals, it can classify the road conditions automatically. It helps the driver's cognitive load and it helps him to save his time for other productive works. And the whole idea of machine learning algorithms is for improving the better living of humankind. So, what is machine learning in a very basic natural it is about teaching computers to take decision on new tasks without human intervention or without writing a code. So, how it does it learns the algorithm based on historical data. So, the computer should take its own decisions when the new task comes without human's intervention how to take the intervention it is based on historical data the computer collected or the human provided or human created the algorithmic. The goal of machine learning or AI is developing machines that can work like humans that can mimic human intelligence and that can do all activities a human can do. By now you might know by seeing this color in this slide there is activity. So, I do not need to tell that you have to pass the video and write down answers and resume to continue. Whenever you see this color slide you have to pass the video do the activity then resume to continue. So, here first why we are to understand machine learning concepts is important for learning analytics. I mentioned that machine learning is everywhere and it is helping humans to live better. So, why we have to understand this machine learning concepts in LA please write down your answers. So, the let us go with the why we have to model the learners interaction behavior we saw the learner interaction behavior in a different environment we can collect data, but we have to model the learners interaction behavior in order to model that we need to know what is machine learning algorithms can be used. And in order to predict the outcomes, suppose given actions you want to predict what will be the students performance in the mid-sem or in end semester exam. So, how do you create model to predict the outcome that is also needed. Also if you know the student is not going to pass the exam in the next mid-sem or midterm how do you provide recommendations what recommendation should work. So, in order to do this we need to know what is machine learning. But why we need to understand the concepts in machine learning why I cannot simply use the existing machine learning tools and apply it why I need to know what are the algorithms why I should know what is machine learning concepts. The reason is we should know how to select a right algorithm for your problem and to choose a correct parameters or to adjust or tune the variables in the algorithms to get a better performance on your algorithm. Also for you to understand the model you developed. Some cases the machine learning algo or model looks like a black box. But if you do not know how the machine learning model able to predict or classify or able to recommend your student if you do not know the information about it you would not be able to use it effectively. So, you have to understand how the model works. I mentioned that you need to understand machine learning concept for the first two reasons that you have to select algorithm or select a correct variable. However, in a current scenario some tools like Big ML or Auto ML which helps you to do that. Given a data it fits into all the available tools or all the algorithms you select and it have a grid search algorithm to change the variable values or the particular grid and able to tell you which algorithm with which parameter works for your data. So, currently the applications help you to do that. But in order to understand how the model works why the model predicted a student as a low score or high score you should understand the machine learning concepts. So, machine learning in education. Machine learning techniques have been used for predicting students performance before they even appear to the exams. So, that we can provide recommendations and we can help the students to pass the exam. Machine learning has been used in an intelligent tutoring system. So, personalized adaptive learning environments that is very common nowadays based on the students response, the next set of actions or content or questions has been created. This machine learning algorithms has been used. Also machine learning is used for essay grading. The students subjects they submit essays or topics automatically grading their essays by using applying the natural language processing techniques. And machine learning also can be used to predict the students' affective states like a student is bored, confused in doing the online lecture or doing the classroom environment. So, I am giving the examples in a different scales. One is using log data that is you can predict their performance. You can use NLP to predict the essay gradings or you can use the facial expressions and log data and some other sensor data to predict the affective states. Also, if you want to predict the students drop out in a MOOC, you can use the log data to use machine learning to predict which student will drop out in which week or which students in the risk of dropping out of the course and much more for machine learning in education. So, we had to use machine learning. When a solution needs to be personalized, for example, in the interchange introducing systems, we need a solution for each an individual student or at least some cluster of students. Now, we can use the machine learning algorithm because we have not seen the students behavior before they coming into a learning environment. We have seen the similar students' performance in the environment. So, if the solution you are going to provide is personalized for students, machine learning algorithm can be used. Like each learner is different in terms of cognition, in terms of the learning ability, the skills, the motivation. So, they are making a recommendation based on each different set of users, you might use a machine learning algorithm. Also, where human expertise does not exist. For example, how human cognition works? Till we do not know exactly one model, say this is how the cognition works. And that model fits for every student, for every subject, no, that is not possible. So, if you do not know how this human cognition works for this particular subject on this particular interval time or if you given this kind of example, it is not possible for humans to create a model for that. So, maybe we can use machine learning algorithms to provide a data or to get the data from the students interaction and provide a recommendation or feedback. And when the rules are difficult to extract, we can apply the machine learning algorithms. Like it is not easy to create a model by manual intervention or manual looking at the data. For example, if the student, if you want to compare the student's attendance versus performance, if you have only two variables like attendance and performance, you might be able to say if the student's attendance is more than 70% you might pass or you might get more than 80 marks. That it comes from your experience also you can create that hypothesis and test it that kind of rules is possible if else rule or you can create a filtering rules. But if the variables is too much and the human cannot create hypothesis or create your own rules from the data, then machine learning can be used. For example, in the learning context, it is very difficult to come up with the rules to detect effective states. For example, when the student will be bored. So student boredom or affect can be visible in terms of the facial expressions, the way they speak or the language they use for writing or the sensors data from physiological sensors. So, how do you create a model to predict student's emotions? Human can do that. When we talk to someone, we can observe their gesture, their facial expressions, their tone. We can guess or we can say which emotion is students under easy anger, easy sad, easy bored, confused. But can we create a model and apply it to predict 60 students in the class? We want to remove the human from the loop because human cannot go and do observations by each and every student and provide a feedback. Can we remove that and can machine can predict the student's emotions so that the machine can give the recommendations. So, if you want to train the machine, then you have to collect a lot of data. There the machine learning algorithms can be helpful. So, there will be a big question, what is the difference between machine learning, AI and deep learning? So, I want to use this slide to clarify that doubt if you have. So, in general, artificial intelligence is something we want machine to mimic like a human. So, you want to create the intelligence which works like a human. And machine learning is part of it. There are many other methods to do that. So, machine learning is one of it. In a machine learning, there are a lot of algorithms like decision tree or probabilistic algorithms or statistical algorithms. We have trees, we have clusters or we have regression models or NAI base. But one of the algorithm is deep learning, that is neural network. So, neural network is the subset of machine learning. Machine learning includes lot more algorithms other than neural network. And machine learning can be considered as a subset of artificial intelligence. Some may not agree with the machine learning versus artificial intelligence subset and set. It is a debate, it is still going on. Let us consider for this course that machine learning is subset of artificial intelligence. And neural network is a subset of machine learning. And in this course, we will not deal much about artificial intelligence, other algorithms or the deep learning. We will talk about some of the machine learning concepts. So, in this video, we saw what is machine learning and how machine learning can be used for learning analytics. We will talk more about machine learning in a subsequent videos. You might be expecting a definition of machine learning because we did not give any definition for machine learning. If you are expecting a definition for machine learning, I recommend you to watch the introduction videos of machine learning course by Andrew NG. It is offered in a course era, but this course is available in YouTube if you do not want to register course in course era. So, please watch the Andrew NG's introduction videos of machine learning, where he gives the definitions and explains. We will talk about that most of his course content in this our course, but we will talk about a bit. We recommend you to go and watch this course in YouTube. Thank you.