ai-reading-car-analysis

เจ้าของผลงานเล่าไว้แบบนี้

https://autoscope-thailand.vercel.app # ai-reading-car-analysis Monthly Thailand new-car registration analysis pipeline (DLT data) and Next.js frontend dashboard. ## Use the dashboard The repository includes the latest generated public JSON, so the dashboard can run without the private/raw Excel inputs. For a non-technical user, double-click `START.bat` at the repo root — it installs anything missing on first run, then opens the dashboard. Nothing else to type or configure. Technical users can run the same two steps by hand instead: ```powershell INSTALL_FROM_ZERO.bat frontend\RUN.bat ``` Open `http://localhost:3001`. The static site is also deployable through the included GitHub Pages workflow. ## Directory Structure - `backend/`: The data processing pipeline (Python). - `frontend/`: The Next.js dashboard application. - `docs/`: Project/product docs (`PRODUCT.md`, `DESIGN.md`, `SKILLS.md`). - `handoffs/`: Current operator runbook and release handoff. For admin operation, file safety, monthly update steps, and what can/cannot be deleted, see `docs/OPERATOR_ADMIN_MANUAL.md`. For a very simple Thai training deck, use `handoffs/thai-simple-admin-manual-2026-07-31.pptx`. ## Pipeline (Backend) The two DLT sources remain at separate grains. Model rows never receive `Powertrain`; fuel rows derive it from `backend/config/powertrain_map.csv`. Model maintenance is CSV-first: - `backend/config/model_map.csv` maps `(brand2, raw_model)` to the canonical `model2` name. - `backend/config/model_powertrain_review.csv`

โดย fietao · ดูโพสต์ต้นทาง