tool for exploring regularly sampled time series data - rajivsam/tseda See https://rajivsam.github.io/r2ds-blog/posts/tseda%20announcement/ for a summary of features.

Lightfield — AI-native CRM that builds itself and does work for you

AI-native CRM that builds itself and does work for you

"Hi everyone! I’m Rajiv, the maker of tseda.

I built this because I saw too many data scientists (myself included) spending hours manually squinting at noisy time-series charts or guessing at seasonal window sizes. We needed a tool that wasn't just another plotting library, but a mathematical filter that could tell us why the data is moving.

tseda handles the heavy lifting of SSA decomposition and change-point detection so you can focus on the 'What's next?' instead of the 'What's this?'. It’s open-source and ready for your toughest datasets.

I’d love to hear your feedback on our suitability gating and how you're using SSA in your current workflows!"

tseda is the first automated Exploratory Data Analysis (EDA) tool designed specifically to bridge the gap between raw time-series noise and actionable business intelligence. While most tools just plot lines, tseda uses advanced Singular Spectrum Analysis (SSA) to surgically separate your data into its true components: Trend, Seasonality, and Residuals.

Why it’s a game-changer:

The Guardrail: It automatically checks if your data is "model-ready" using a Durbin-Watson-backed suitability gate, preventing you from building on pure noise.
The Heuristic: It eliminates manual tuning by using an AIC-based model selection to find the perfect decomposition window every time.
The Workflow: A three-step "Explore, Decompose, Export" framework that moves from raw CSV to validated signal in under 5 minutes.
Unified Interface: A high-fidelity Dash UI for visual discovery that stays in 1:1 parity with its Python API, so your research script always matches your dashboard.

The repo: https://github.com/rajivsam/tseda has plenty of examples.