 Welcome back. This video will talk about the main factors that data collection, there is no data, how to collect data, where to get data and what is the research community. So after learning this course, what is NEST? So NEST is you can apply learning analytics in your context, if you are a teacher please apply it. If you are a student, you can apply learning analytics if you have collected data something and you do data collection testing, that is the next thing to do and that is if your interest is towards doing a research in LA, collect data, different research questions and do testing and apply it and write published papers, that is what I want, I would be happy if you do that. And if no I am not a researcher, I just joined this course to know what is the basics in ML algorithms, what is data means. So to go and explore advanced topics, I was telling in last video that you can go to each video and talk about, we will talk about resources or you can find your own resources in the internet and read it. If you are interested in the tools or data collection, you are looking at a job in an entry level jobs in a data science career. I would recommend you to go and learn more tools, apply ML in different domains and create a, you know, a password tree where you have applied data, you created models and maybe create a GitHub libraries or create a web page where you know, you apply it, what are you learned and you will then apply for a job then that might be easier to get a job. So that is the whole idea of three directions you can go. One is I want just to know more about machine learning, three topics I want to do the such, do the such and you want to looking for data science job, don't just stop learning domain, apply for other domains or more about tools, this is all interesting. I will talk about what tools is industry use and everything, how to find it out, I will talk about that in the next video. In this video, I will talk about research kind of flavor in L.A. So the general question everybody asks is I do not have data, I do not have system, I am not expert in writing scripts, I do not know data, which data to collect, where to collect data, there are plenty of data already available in online like a Kaggle or a data shop, everything is available, but that will not answer the research question you are looking for. If you are trying to answer a particular such question, I would recommend you create your own system and collect data. But yeah, not everybody is a computer science engineer, I do not have system to collect data, that is perfectly all right. The only thing is due to advance in computer science field, there is no need to be that you write to no programming to create the environment, the things are changing. In the other end, and I see people, a lot of students, especially third year and fourth year students, third years or final year students in different departments, engineering or science, they always come and ask you looking for, I want to work in a real problem, computer science, mechanical, civil, electronic, it does not matter which department, you know, visual communication, they all want to learn programming and they want to create something. So, if you have access to students, give this as a challenge, give this as a work, ask the students to create a system, ask the students to create a front end back end system, maybe Django with Python or Node.js, Ruby on Rails, something like that, ask the students to create a system where server client communications are already established. And ask the students to observe all the clickstream data in the client, store it in a MongoDB or any structured SQL. The students understand all those things and ask them to create, the moment you ask them to create this kind of data, they will be very happy to do it because it is a real world problem. I see a lot of students come to me saying that I want the data, I want to do something in machine learning in a tech, I want to do. And they look for, for me it is a data. I said no, I do not give data to anyone. The reason is it is the data I collected based on my understanding and I can give some data, but it do not serve a purpose. If you, if I give a data to you and you click something, few clicks in the machine learning tool and you create something, that is not going to be any learning. Anyone can do that, that we do not need a complete understanding of what is each mode and everything. For a research, you have to come up with your research question and what data to collect and how to use the data to find, answer your research question. So, look for a people, students especially who are third and fourth year students who are, who are I know, they are interested to create work on a real problem, give them this as a problem and they will help you. If you are a student already, you are looking for it, take this as a challenge and create a system, check for Django Python or Node.js or anything which is latest and recent and try to create a system where server client communication happens, all the client's clickstream data like watching video, reading page, everything clicks has been copied, moved to the server, located, then test it with your network and if it works, upload it in some digital clouds, the free clouds available for students, for the student ID, they will see some free clouds available, use those services, upload it and conduct study. This is how you have to start. If you say I want to start only with the data and extract features, that is not really good. If you know I created a system, I collected data, I answered this question, the prospect increases and your job market also is there a lot of for you. If you are a researcher, you want to looking for this data, you are creating a new system, you are contributing something new to the community saying that this is our system and your name comes up, there is a system created and there is a data collected and there is a study done. So, you are establishing a research network using your system. So, I always recommend go for that. But no, I do not care about that, I just want to apply ML algorithm on XYZ data. That is not learning analytics especially, I was thinking you need to know what data to collect, but you do not want that, I just want to do that. I would say go to online resources. Learning analytics is more about which data to collect why and write the simple scripts in the very simple way. What you can do, you can create a Google site. It is easy to create, there is no programming knowledge needed to create a Google sites. You create multiple pages and you can add a video and you can ask the students to go and take a quiz and you can write a simple small script in the Google which captures all the students clicks. I would recommend to go and check that, how to do that, ask the students to do it, it is all you can start with that. So, if you are interested in one particular environment, say we, I mean I have just looked at this paper, this environment is available, I want to use it, but contact the authors, they might be happy to share, some will be happy to share. If it requires a lot of training and onboarding process, they will not do it, but if it is available freely online, talk to the researcher, researcher will be happy to share. As long as you acknowledge them and you inform them what are you doing with that, you are looking for data, they will be happy to share. So, contact those authors, they will be happy to share, but make sure that what you want and what is the data and talk to them in detail. Simply, I want this system will not help. You have to give a detail why, why did you read it, what you are planning to do, give those interesting, you were kind of some synthesis on that interest toward that particular environment which you read in a particular research paper, go and talk to the authors. That is the one way you can do, I actually do that with couple of authors who created environments because I do not myself create environments as such as of now. And more about after that you should know what is future extraction and models from the research questions and solving it is what learning analytics is not just applying ML on XYZ data. So, I do not care about all, I just want a data. Go to data shop, Penn State Learning Center created something called data shop just hosted in a CMU site. Go there, look at the data, register yourself. You have to contribute a data, but fine if you are not contributing you can access to a basics data, look at the data shop, the lot of data from all the resources, most of resources in academic learning instruments uploaded there can use it. And the thing is the data available to use that you will not get more data if you want to do another kind of research, you will not be able to do it, but with that data you have to live with that and you have to answer such question on that. Other good resources Kaggle, Kaggle now and then even now is they always upload up 100,000 of users data, click streams, lot of data has been uploaded in Kaggle and that Kaggle data is also useful. You can download the data, you can ask a different research question from the what asked in the Kaggle. So, yeah do that, but Kaggle may not allow all the data can be freely used and for such purpose. So, check though that there are the constraints and talk to the developer if you are interested in analyzing the data for some other thing, something else, get the permission. That will be needed for the publications. If publisher find out this data is not yours and Kaggle and they will check whether you got permission from Kaggle or not. We talked about in the last video also no, we did not talk about linear regression, i-bay relation to these three, I did not talk in detail on logic regression. So, I again repeat that go and watch the videos. I request you to watch Andrew Andrews video to understand what is this three things. If not, any resource talks about machine learning, ip parameters books or videos will be good. And now come to the research areas in LA. LA is just a topic, what I was trying to say is a collect data, apply some machine learning. What is the areas I want to do? The basic thing everybody does is modeling learners performance, modeling learners engagement, modeling learners skill in environment, modeling learners to predict the performance, to mark something like that. Or engagement is something new and it has been for some time, but people are working newly on engagement to say how multiple environments is getting the engagement of student in one place. Our content analytics, education is all about content. You are giving a video, you are giving about test, not just the interaction, but also the content. So, a lot of people do on content analytics, just check the content also. You can apply machine learning and you can apply pattern mining algorithms and you extract the strategies based on students interaction data. What are the strategy difference between high performance and slow performance like a DSM, differences you can find, you can do that. A lot of people do research on a privacy, how to store data, what is ethics in privacy also, there are a lot of research papers published on that, look at those papers. I talked about affective computing in a multi-model learning analytics interaction. So, yeah, you can do the multi-model. People also do a data visualization like a dashboard, they are innovatively come up with the new things, how it is useful for something. You can, you saw the ISAT, ISAT is kind of a new direction of the data visualization in the learning analytics community. And last decade, people talk about self-regulated learning and it is advanced topic. It is not just one data, I think all data from multiple channels and multi-model data and you have to analyze that to create a self-regulated learning model. And the advanced topic is, intelligent tutoring systems or personalized adaptive learning environments where you not only predict it and you are prescribing them because just suggesting something new to students to achieve something. And through intelligent learning environment, which should be completely free from any fixed set of rules, not exist the world. Lot of people do ITS, intelligent tutoring system just based on the performance in the current exam, not even the history of performance. The successful ITS all do only the performance based adaptation. So, you can do something on that angle also. How do we know about these research areas from that? How do we get it? I will tell you. What I did, I went and checked the EDM education data mining and LAK groups. These are the conference proceedings available. This is the EDM. This is the EDM conference resource. You can show and check it out. And look at the abstract papers and all this EDM conference papers free for everyone. Look at those abstract papers and check what is the title, from the detail you understand what is the topic I was talking about. Similarly, for LAK, they put all their things in the solos such that they own the community. What I would say is go and look at the paper. If you do not have access to paper, check the title in the Google Scholar and see Google Scholar might give you the PDF access. And if you do not have further, look at the similar papers with the same authors which have access to you. Nowadays, most researchers, most authors upload the papers in their own personal web page. Go and check the personal page or the research kit. It is all happening. People are moving towards more open and open compared to keeping it for a cost. There is no need to pay to a journal unless it is necessarily important. What happens is you can talk to a researcher and you can find out the similar research publications for the same researcher. It will be available. Go and read that. Educational keep, completely open source. Solar search also completely going open source. So, yeah, you can get a lot of open source data. Check papers in these communities. Other conferences which are really interesting is AIED, Artificial Intelligence and Education. It happens every year now. So, look at those conference or Intelligent Tutoring Systems. This also does something in learning and takes modeling students to provide something. These are international community, EDM, LAK, this. And there is something called Asia specific things called for Asian International Conference on Computers in Education. This is by a group 6CSE in Asia. So, check this, not 6CSE, APCSE, so Asia. So, check this website also, International Conference on Computers in Education. And each of these guys have their own associated journal, IJED, RPTEL, JEDM Journal of EDM, LAK Journal, or there are something called I-Called. There is a good conference called I-Called. This is also international. They have a conference on I-Tribally Transactions called TELT or I-Tribally Transactions on FIT Computing AC. Transactions on FIT Computing, Transactions on Learning Technology. There are a lot of journals, associated journals. When you look at this conference, you might see get the associated journal names. Look at those papers and get the research field. If you are looking for which research topic to start in LA or if you are looking for which topic to start in LA, I would request you to go and read last two years conference proceedings of EDM and LAK, not whole proceedings. Just look at the title and abstract. If you by reading it, you will come to know various your interests. Last three years, 2018, 19, 23 years is a good time to start. And look at it last three years, conference abstract and proceedings. You will understand what is the topic is about. Then pick the topic which you are interested in and topic which you can do the search on, which topics which you can collect data, we have access to those resources. Depending on that, choose the topic. These are some other good interesting conferences where you can publish your work. So, that is about the LA community. In India, we have, we run this conference called ICC, I-Tribally Conference on Technology of Education. It is mostly technology, but we also do a separate stream for that learning analytics in it. So, if you are interested in Indian conference, look for this T4E and check those papers. It is all open access after one year. So, check those papers. Also, ACM group in India conduct a conference on computer education to check those papers also. And this is a lot of repeated research labs in India. For example, IBM or V Pro or Adobe, they publish papers in international forum. It is not that there is no education research or learning analytics is not happening in India. There are a lot of good companies for education initiatives, do research in this area. All the commercial companies also do research in this area and publish papers in international forum. So, look at those companies articles if it is available online. Thank you.