 In this learning dialogue, we will talk about types of learning analytics. Learning analytics is also defined as analytics applied on learning data. As we see discussed in the first video, learning analytics definition is not fixed or is still ever being. So, there are few researchers, we consider that learning analytics is simply the analytics applied on learner data. Hence, we look at learning analytics in that lens, we can apply types of data analytics in LA also. So, we can apply the types of data analytics in LA, there can be four types that is descriptive, diagnostic, predictive and prescriptive. Descriptive is actually to shows what happened. If you are lot of data, you have to report the data in a chart, I only put in a pie chart, you want to report a data in the line graph or what is happened, you want to look at the data. You can write a report a data in the text form also, it is about reporting the data like you are collected too many data, you have to report the data, what happened in the data. The next step is diagnostic. In diagnostic analytics, we want to apply why this particular thing happened, you have a data, there is suddenly some change in the data, you want to show why this particular change happened in the data. In a predictive analytics, now you have the knowledge about why the data happened, what is the data, can be predicted what will happen in the future. So, it is a predictive analytics. The fourth type of an analytics is called prescriptive analytics. In prescriptive analytics, you want the user to achieve some level, the user currently in the some different level. What kind of information we can provide, what kind of hint, what kind of feedback you can provide. So, that the user can achieve the level you wanted them to go or the user said this will not go, that is called prescriptive analytics. We will learn each type of an analytics by doing a small activity. Let us look at our example again. The example we discussed in our first class, that you are a class teacher and you have access to data for last 5 years such as academic background, profile, performance and also the participation in the course. You have all this data, how will you report this data and what data you will report, think about your answer. You can pass this video, after writing your answer you can resume the video to continue. The descriptive analytics is the first type of an analytics, it is to show the data, report the data using visuals also by the text methods. For example, if you said you want to report the attendance or you want to report the performance over years, yes that can be represented by using some pie charts or histograms or line graphs or some scatter plots or even more interactive visualizations. We might have seen a dashboard in a YouTube or in a Moodle data you might have seen a lot of dashboard. So, this dashboard data is actually called descriptive analytics, where you show the data what has happened. By looking at the data in the visual form, the researcher or teacher can say oh yeah at this particular class the student participation is less or in this particular example the student were not able to answer the question correctly. So, looking at a data the learner can understand or the researcher can understand what happened in the particular course. So, this is actually mostly used in a class level also in the academic analytics and this dashboard is presented for the stakeholders for example your customer or your students of a teacher also to discuss with the other researchers. Here are a couple of examples the standard examples of reporting the visuals like a pie chart or the bar chart in this pie chart you saw the distribution of male participants as a female participants there are more male participants in this class. And there are past percentage over the last five years of data like what happened in last 2018-17 to 2014 like the past percentage of a particular class in a particular course maybe your course you are teaching for last five years. By using this data you can analyze that in the 2015 maybe the question paper was tough the question paper contains some topics which you are not teaching in the class. Now you might go and check that question paper from 2015 and look at what are the contents you thought if you are not thought some topics in that you can give a more thoughts about it or the questions were not asked the way you thought you can reconsider that. So by looking at a data you can understand what happened over the years or what happening in the class and what is happened in the data. The next type of analytics is called diagnostics is next level diagnostic analytics includes descriptive also analyzing the data to answer why X happened why something happened it is called diagnostic analytics the X can be related to learners performance or learners participation and anything let us do the activity to understand more about diagnostic analytics you have collected a data from your past courses and also you represented them as a bar charts line graphs or pie chart as a course instructor what you would like to analyze from the data you are the course instructor you have collected data you represented using some visuals and what you want to do that is what question you would like to answer from the data you have any questions you can pass this video and write your answers after completing your activity you can resume this video diagnostic analytics analysis the data to identify what something has happened why it is happened for example why a student drop out of the course or why a student fail in an exam or why this particular student was not able to perform this particular question in this exam you can go such a fine-grained analysis or how many students has passed in midterm such a high-level analysis also can be possible the preprocessing methods has been discussed mostly in other ML courses we will give the links to those courses when we talk talk about this diagnostic analytics so the first the raw data is processed then the relevant features from this data is extracted for the analysis these features are used to find what something happened why it is happened there are very few techniques applied for diagnostic analytics such as pattern mining correlation or regression techniques are used to do why it is happened the correlation analysis there are four correlation analysis chart is shown here in the upper left chart you can see the correlation is very medium it is not highly correlated however in a lower left chart you can see the correlation is I but the data leads to different correlation now so this kind of correlation analysis tells when x happens what happens to y when x increases y also increases