Orange: Classification Tree
Sumber: https://orange.biolab.si/workflows/
Workflow ini menggabungkan interface dan visualisasi classification tree dengan scatter plot. Ketika keduanya tree viewer dan scatter plot di buka, pemilihan sembarang node dari tree akan mengirimkan contoh data terkait ke scatter plot. Dalam workflow, data yang dipilih diperlakukan sebagai bagian dari seluruh dataset dan ditayangkan dalam scatter plot. Dengan kombinasi sederhana dari widget, kita telah membangun browser klasifikasi tree interaktif.
The workflow in Orange Data Mining shown in the image follows a Classification Tree approach, allowing for interactive decision tree exploration and visualization. Here’s the step-by-step breakdown:
1. File (Data Loading)
- The File widget loads the Iris dataset ("iris.tab"), which is a preloaded documentation dataset.
- The dataset consists of 150 instances belonging to three classes of the Iris flower species.
2. Classification Tree (Building the Model)
- The Classification Tree widget is used to train a decision tree model based on the Iris dataset.
- The model splits the data into nodes based on feature importance, creating a tree structure.
3. Classification Tree Viewer (Interactive Tree Exploration)
- The Classification Tree Viewer allows users to double-click on any tree node to inspect its data.
- Users can explore different branches of the tree to understand how the model makes classification decisions.
4. Scatter Plot (Visualizing Data Selection)
- The Scatter Plot widget is used to visualize how different features influence the classification.
- Any change in tree node selection (from the Classification Tree Viewer) updates the scatter plot, helping users analyze the separability of classes.
5. Box Plot (Analyzing Feature Distributions)
- The Box Plot widget helps in analyzing feature distributions within selected tree nodes.
- The data selected in the Classification Tree Viewer propagates to the Box Plot for further statistical analysis.
Interactive Selection and Propagation
- Any change in tree node selection dynamically updates the Scatter Plot and Box Plot widgets.
- This makes the workflow interactive, allowing users to inspect different decision boundaries.
Summary This Orange Data Mining workflow loads the Iris dataset, builds a decision tree, allows for interactive tree exploration, and visualizes classification results using scatter plots and box plots. It enables interactive decision-making analysis, making it ideal for understanding classification models and their impact on the dataset.