 In all my previous videos in the Introduction to Data Science series, Orange Widgets sends data instances to each other. For example, we loaded Human Development Index data with the Datasets widget and then looked at the data in the data table. Notice that Orange reports that the two widgets communicate using the data channel. We also used the scatter plot and the output from the Datasets widget is again labeled as data. But the information carried from one widget to another does not necessarily contain only the data instances. For example, Orange widgets can also output distance matrices, objects that learn from the data, evaluation results or a set of data features. Okay, I'm going too fast and we will learn about these types of channels in our later videos. But let me make an example here. I will use a correlations widget that shows, well, correlations, between two features in my dataset. The widget sorts the feature pairs by the absolute correlation. We can see that some features in Human Development Index data are strongly correlated. For instance, countries with high infant mortality rates also have high child mortality rates. More interestingly and besides the most apparent correlations, inequality of education is negatively correlated with mean years of schooling. I would like to see this particular relation in the scatter plot. I could open a scatter plot and manually find this specific pair of features. But I would need to do this for every pair I find interesting in the correlations widget and I am too lazy to do this. Instead, let me connect correlations to the scatter plot. Orange asks me what type of connection I would prefer. I do not want correlations to emit the data with only the chosen set of two features. So I click on the link and cancel this signal. Instead, I see that correlations also outputs the information on the two features I can select in the widget. The scatter plot can receive this information on the features channel. Let me establish this communication channel by dragging a line from correlations features box to the scatter plots features box. Nice. I see that orange now adds the label features on the link between correlations and scatter plot. I will now open both widgets. Selecting a row in correlations now emits a signal that informs the scatter plot which axis to display. I see why education inequality is negatively correlated with mean years of schooling or why there is such a high correlation between male and female populations with at least some secondary education. In this video, I have introduced a widget that computes correlation between pairs of data features and use it to demonstrate another type of channel. In my next video, I would like to close my introduction to orange workflows and visual programming by showing you the widgets that report on the distribution of feature values.