 Creating a data analysis workflow in Orange is simple. We start with opening the data in the file widget. Some preloaded datasets are available, so let's select RS and explore it. Let's check out the data in a data table. I will select the data table widget and connect the file widget to it. Here we have 150 RS flowers from the famous Fishes dataset. Flowers are described by four features, the length and width of saples and the length and width of petals. Each flower is labelled with one of the three classes, a species of RS, RS etosa, RS versicolor and RS virginica. Now let's visualise the data. I'll connect distributions widget to the file widget. In this way, the file widget sends any data it loads to the distributions widget. In distributions, we can walk through all the features in the data. Petal length and petal width seem to nicely separate different species of RS. We can additionally inspect the data in the scatter plot. The plot that we see is a bit messy. RS versicolor in red and virginica in green are not well separated. I wonder if there's any pair of features that would nicely separate the three classes. I can click rank projections to score or feature pairs. A higher score indicates a better separation of different species of RS. The best score scatter plot with petal length and petal width really nicely separates data instances of different class. But there's some overlap of RS versicolor and RS virginica. I'll select data instances in the overlapping region. Scatter plot widget automatically sends the data to its output. Now I connect another data table to the scatter plot to inspect selected data instances. No surprises here. They are all either RS versicolor or RS virginica and seem to have similar values for all four features. We can expand the workflow with other widgets or save it for frequent use. But for now that's it. We've learned that orange widgets communicate with one another and the changes in one widget are immediately propagated through the workflow. End of workflow.