We assume you have the default Orange installation and have not changed widget defaults. Most figures are upgrades of previous figures, so you should follow the commands from the beginning. All network widgets are in the Network widget tab. - run Orange Canvas Figure 1 - add File widget (Data tab), double click to open it - click on Data File combo box, select "Browse documentation data sets...", and select iris.tab file - add Select Data widget (Data tab) - connect File and Select Data widgets - double click Select Data widget to open it - select Attribute: iris, Operator: in, Values: Iris-versicolor and Iris-virginica, and click add - add Test Learners widget (Evaluate tab) - connect Select Data and Test Learners widgets - add Logistic Regression widget (Classify tab) - connect Logistic Regression and Test Learners widgets - add k Nearest Neighbours widget (Classify tab) - connect k Nearest Neighbours and Test Learners widgets - add Classification Tree widget (Classify tab) - connect Classification Tree and Test Learners widgets - double click Logistic Regression widget to open it - double click Test Learners widget to open it - select Performance scores: Classification accuracy and Area under ROC curve Figure 2 - add Scatterplot widget (Visualize tab) - connect File and Scatterplot widgets - add Test Learners widget (Evaluate tab) - add Confusion Matrix widget (Evaluate tab) - connect Test Learners and Confusion Matrix widgets - add Scatterplot widget (Visualize tab) - connect Select Data and Scatterplot widgets - connect Confusion Matrix and Scatterplot widgets - open Scatterplot and Confusion Matrix widgets, select one of the cells in Confusion Matrix and look at the resulting image in the Scatterplot - connect Test Learners widget with Scatterplot, Classification Tree, k Nearest Neighbours, and Logistic Regression - select data in Scatterplot and inspect how classifiers are learned on the visually selected data subset Figure 3 - click File/New Scheme - add Net File widget, double click to open it - click on Graph File combo box, select "Browse documentation networks...", and select pubmed.gpickle file - add Pubmed Network View widget - add Net Explorer widget - connect Pubmed Network View and Net Explorer widgets - connect Net File and Net Explorer widgets - open Pubmed Network View - select the first paper in the Paper Selection list - open Net Explorer - select Optimize: Method: Fruchterman Reingold, Iterations: 1000 - click Optimize layout - select Node Labels: title Figure 4 - click File/New Scheme - add Net File widget, double click to open it - click on Graph File combo box, select "Browse documentation networks...", and select lastfm.net file - add Scatterplot widget (Visualize tab) - connect Net File and Scatterplot widgets - add Net Explorer widget - connect Net File and Net Explorer widgets - add Scatterplot widget (Visualize tab) - connect Net Explorer and Scatterplot widgets, connect Selected Items output - add Net Explorer widget - connect Net Explorer and Net Explorer widgets Figure 5 - open the first Net Explorer widget - select Node labels: artist - click Mark tab, enter Paul Simon in the Search box - click Marked to selection button (at the bottom) - select Mark: Neighbours of selected - click Nodes tab - click Show labels on marked nodes only - click Zoom button - zoom-in the region with marked nodes (drag the mouse to zoom into rectangle) - click Marked to selection button again - click Send selection button - open the second Net Explorer Figure 6 - click clear selection button - select Node color atribute: best tag - select tooltips: best tag - hover over nodes to observe the genre Figure 7 - click File/New Scheme - add Net File widget, double click to open it - click on Graph File combo box, select "Browse documentation networks...", and select lastfm.net file - add Net Explorer widget - connect Net File and Net Explorer widgets - add Select Attributes widget (Data tab) - connect Net File and Select Attributes widgets - add Select Data widget (Data tab) - connect Select Attributes and Select Data widgets - connect Select Data and Net Explorer widgets, connect Matching Data with Items subset Signals Figure 8 - open Select Data widget - select Attribute: first album, Operator: "<", Values: 1985 - click Add - open Net Explorer widget - select Node color atribute: best tag - open Select Data widget again - change Values to 1990 - click Modify - view the network in the Net Explorer widget - repeat the last 3 steps for years 1995 and 2000 Figure 9 - right click Select Data widget, select Remove - add k-Means Clustering widget (Unsupervised tab) - connect Select Attributes and k-Means Clustering widgets - open k-Means Clustering widget - enter Clusters: Fixed: 22 - select Distance measures: Manhattan - edit Cluster IDs: Name: rename to genre - connect k-Means Clustering and Net Explorer widgets, connect Data and Items signals - open Net Explorer widget - select Node color attribute: genre - click on Mark tab, select Mark: None Figure 10 - click on Nodes tab - select Node size attribute: num of albums Figure 11 - right click the connection between k-Means Clustering and Net Explorer, and select Remove - add Select Attributes widget - connect k-Means Clustering widgets and Select Attributes widgets - connect Select Attributes and Net Explorer widgets - add Scatterplot widget - connect Select Attributes and Scatterplot widgets - connect Net Explorer and Scatterplot widgets - add Data Table widget - connect Net Explorer and Data Table widgets - open Net Explorer, click on Mark tab, select Mark: most connections: 50 - click Marked to selection - click Send selection - open Scatterplot widget - select X-axis attribute: genre - select Y-axis attribute: plays Figure 12 - repeat steps from Figure 1 - add Confusion Matrix widget (Evaluate tab) - connect Test Learners and Confusion Matrix widgets - add Example Distance widget (Unsupervised tab) - connect Select Data and Example Distance widgets - add Net from Distances widget - connect Example Distance and Net from Distances widgets - add Net Explorer widget - connect Net from Distances and Net Explorer widget, connect Selected Data and Item Subset signals - open k Nearest Neighbours widget - select Number of neighbours: 3 - open Net from Distances widget - select kNN: 3 - open Confusion Matrix - select Learners: kNN - select one of the cells in the prediction matrix - observer mis-classified examples marked in the Net Explorer Figure 13 - click File/New Scheme - add Net File widget, double click to open it - click on Graph File combo box, select "Browse documentation networks...", and select airtraffic.net file - add Net Clustering widget - connect Net File and Net Clustering widgets - open Net Clustering widget - check Commit automatically and click Commit button - add Net Analysis widget - connect Net Clustering and Net Analysis widgets - open Net Analysis widget - select all Graph-level and Node-level indices except: Radius, Number of weakly connected components, In-degree, Out-degree, In-degree centrality, Out-degree centrality, Random-walk betweenness centrality, Approx. random-walk betweenness centrality, Eigenvector centrality (Numpy), and Closeness vitality - check Commit automatically and click Commit button - add Net Explorer widget - connect Net Analysis and Net Explorer widgets - add Select Attributes widget - open Select Attributes and select only node-level index attributes attributes (computed in the Net Analysis widget) - connect Net Analysis and Select Attributes widgets - add Rank widget (Data tab) - connect Select Attributes and Rank widgets - sort by ReliefF and select Best ranked: 7 - add Test Learners widget - connect Rank and Test Learners widgets - add Logistic Regression widget - connect Logistic Regression and Test Learners widgets - add Random Forest widget - connect Random Forest and Test Learners widgets - add Confusion Matrix widget - connect Test Learners and Confusion Matrix widgets - open Confusion Matrix - select Learners: Logistic Regression - select all mis-classified examples in the Prediction matrix - connect Confusion Matrix and Net Explorer widgets, connect Selected Data and Item Subset signals - add Data Table widget - connect Net Explorer and Data Table widgets Figure 14 - open Net Explorer widget - select Node color attribute: clustering label propagation - select Node size attribute: passengers Figure 15 - open Net Analysis widget