A SwiftUI macOS application that demonstrates how a perceptron works for binary classification. Perfect for educational purposes and conference demos!
Note: The initial version of this app was entirely vibe coded using Claude Code. I didn't write a single line of code by hand.
- Real-time decision boundary: Watch the linear separator update as you adjust weights and bias
- Smart chart scaling: Automatically fits each dataset perfectly, no jarring scale jumps
- Properly clipped decision line: Boundary line stays within chart bounds
- Dynamic labeling: Axis labels, legends, and equations automatically update based on CSV headers
- Self-documenting charts: Each dataset shows its own meaningful labels (e.g., "Sepal Length" vs "Sepal Width")
Choose from preset datasets via the dropdown menu:
- AND Gate - Classic boolean logic gate
- OR Gate - Another fundamental logic gate
- Iris Flowers - Setosa vs Versicolor classification (sepal dimensions)
- UIKit vs SwiftUI - Humorous developer dataset comparing:
- X-axis: Development time (hours)
- Y-axis: Number of bugs
- Labels: UIKit (slower development, fewer bugs) vs SwiftUI (faster development, more bugs)
- Weight sliders (W1, W2): Range -10 to +10 with 0.1 step precision
- Bias slider: Range -10 to +10 with 0.1 step precision
- Snap-to values: All sliders snap to 0.1 increments for precise control
- Live equation display: Shows the current decision boundary equation
- CSV text editor: Paste your own training data
- Smart header detection: First line can contain column headers for automatic labeling
- Flexible format:
header1,header2,output_labelor just data rows - Parse button: Load your custom dataset
- Real-time feedback: Shows number of parsed data points
- Reset View button: Instantly fit the chart to show all data points
- Double-tap chart: Alternative way to reset the view to fit data
- Built with Swift 6: Full concurrency compliance, no warnings
- Native SwiftUI Charts: Uses built-in Charts framework for optimal performance
- macOS Application: Designed specifically for macOS with proper window management
- Educational Focus: Clear visualizations and interactive controls for learning
- Select a dataset from the dropdown menu
- Adjust the weights (W1, W2) using the sliders
- Modify the bias to see how it shifts the decision boundary
- Observe how the green line separates the two classes of data points
- Experiment with your own CSV data by pasting it in the text area
The perceptron implements a linear decision boundary defined by:
W1 × x₁ + W2 × x₂ + bias = 0
Points above the line are classified as one class, points below as another.
- Open
PerceptronViz.xcodeprojin Xcode - Build and run (⌘R)
- Requires macOS 14.0+ and Xcode 16+
The UIKit vs SwiftUI dataset is specifically designed as a humorous talking point for iOS developer conferences, playfully highlighting the trade-offs between development speed and bug frequency in different UI frameworks.
Built with ❤️ using SwiftUI and Swift Charts