Install the package via your preferred manager:
uv add trendifyor with pip:
pip install trendify- Built to scale: process thousands of runs without holding every run's data in memory at once. Throughput stays flat whether you have dozens of runs or tens of thousands.
- Typed records, no migrations: points, traces, tables, and histograms are validated Pydantic models. Add your own record types anytime without writing a schema migration.
- Parallelizable:
trendify generatecan fan your processing function out across multiple CPU cores with--n-procs, with multiprocessing-safe logging that funnels every worker's output through a single queue. - Static assets or a live dashboard: render tagged data straight to Matplotlib images and CSV tables with
trendify render, or launch an interactive FastAPI dashboard withtrendify viewerto browse tags, tables, and interactive plots in the browser.
- Scalable. Throughput stays flat whether you're processing dozens of runs or tens of thousands.
- Memory efficient. Each run is processed once and cached on disk, nothing needs to stay open or held in memory for the whole sweep. This is critical for processing large amounts of data with less memory than is available when batch processing.
- Flexible output. Render static Matplotlib images and CSV tables for a report, or browse the same data interactively in a live dashboard.