PlotOxide

A high-performance data visualization and Statistical Process Control (SPC) application built with Rust and egui.

Features

Data Import

• CSV and Parquet file support via Polars
• Automatic type detection and timestamp parsing
• Multiple Y-series support with interactive selection
• Drag-and-drop file loading

Visualization Modes

• Scatter/Line plots with configurable styles
• Histograms with adjustable bin count
• Box Plots for distribution analysis
• Pareto Charts with cumulative percentage line
• X-bar R Charts for subgroup-based SPC
• P-Charts for proportion/attribute data

Statistical Process Control

• Configurable σ limits (1-6σ)
• Sigma zone visualization (±1σ, ±2σ, ±3σ)
• Western Electric rules detection
• Process capability analysis (Cp, Cpk)
• Specification limits (LSL/USL)
• Outlier detection and highlighting

Analysis Tools

• Moving Average (MA) overlay
• Exponentially Weighted Moving Average (EWMA)
• Polynomial regression (linear through 4th order)
• Real-time statistics (mean, median, std dev, min, max)
• Data filtering (X/Y range, outliers, empty values)

User Interface

• Dark/Light themes
• Interactive tooltips and point selection
• Collapsible control panels
• Keyboard shortcuts
• LTTB downsampling for large datasets

Installation

git clone https://github.com/brainmod/plot-oxide.git
cd plot-oxide
cargo build --release

Usage

cargo run --release

1. Open a CSV or Parquet file (📂 button or drag-and-drop)
2. Select X-axis column or use row index
3. Select Y-series from the left panel (Ctrl+click for multi-select)
4. Choose visualization mode and enable SPC features as needed

Keyboard Shortcuts

Key	Action
G	Toggle grid
L	Toggle legend
T	Toggle theme
R	Reset view
H/F1	Help

Architecture

src/
├── main.rs          # Entry point
├── app.rs           # PlotOxide application struct
├── constants.rs     # Configuration constants
├── error.rs         # PlotError type
├── data/
│   ├── source.rs    # DataSource (Polars wrapper)
│   └── stats.rs     # Statistics calculations
├── state/
│   ├── view.rs      # ViewState, PlotMode, LineStyle
│   ├── spc.rs       # SpcConfig, WEViolation
│   ├── filters.rs   # FilterConfig
│   └── ui.rs        # UiState
├── ui/
│   ├── toolbar.rs
│   ├── series_panel.rs
│   ├── plot.rs
│   ├── stats_panel.rs
│   └── data_table.rs
└── widgets/
    ├── spc_controls.rs
    ├── filter_controls.rs
    └── range_input.rs

Technology Stack

Component	Technology
Language	Rust 2024
GUI	egui/eframe
Plotting	egui_plot
Data	Polars v0.46
Serialization	serde
Errors	thiserror

Performance

PlotOxide includes comprehensive performance optimizations:

Benchmarks (100k rows, 1 series selected, all filters/overlays enabled):

Operation	Time	Notes
render_data_table	42ms	Down from 1500ms (35x improvement)
render_plot	9.8ms	During rapid panning
swap_buffers	10ms	GPU present
File load (CSV)	32ms	Via Polars
Stats calculation	2ms	Cached after first compute

Optimization Features

• Column-string prefetch in data table (critical fix: avoids per-cell column conversion)
• LTTB caching with zoom quantization (10-50x fewer recomputes)
• Virtual scrolling for data table (O(visible) instead of O(n))
• Adaptive downsampling (fast nth-point during drag, LTTB when settled)
• Point culling via binary search for viewport optimization
• Background threading for non-blocking file loads
• Shared memory using Arc for zero-copy data sharing
• Stats caching with version-based invalidation

Profiling

# Build with profiling enabled
cargo build --release --features profile-with-puffin

# In another terminal
cargo install puffin_viewer
puffin_viewer

See CLAUDE.md for full development notes and architecture details.

License

MIT License - Copyright (c) 2025 Corey Swinth