Practical data preparation before analysis — without Excel freezing. 在分析之前，先把資料整理到「能用」的狀態，不再被 Excel 卡住。
A lightweight, web-based workbench for previewing, filtering, reducing, and analyzing large CSV/XLS datasets before further processing.
一個輕量級的 Web 資料工作台，用來在進一步分析或處理之前，快速預覽、篩選、縮小與初步分析大型 CSV / XLS 資料集。
When working with large CSV or Excel files, common tools like Excel often freeze or become unusable. Before running any serious analysis, data usually needs to be previewed, filtered, and reduced first.
dataset-workbench is built to solve this exact problem.
It is especially suited for transactional datasets (such as orders, events, or record-based data),
where row-level inspection, filtering, and reduction are required before analysis.
當資料量一大，Excel 很容易卡死。 實務上，在進行任何分析之前，資料通常需要先查看、篩選與縮小。
dataset-workbench 正是為了這個真實痛點而生。
本專案特別適合用於交易型資料集（例如訂單、事件或各類紀錄型資料），
重點放在分析前的列級（row-level）預覽、篩選與資料縮減。
- Preview large CSV / XLS files without loading everything into Excel
- Display schema, row count, and sample rows
- Scrollable preview table with safe layout (no page overflow)
- 不需用 Excel 即可快速預覽大型 CSV / XLS
- 顯示欄位結構、總筆數與前幾筆資料
- 預覽表格具備安全捲動，不會撐爆畫面
- Build filter conditions visually
- Support global AND / OR logic
- Clear UI indicators showing current filter logic
- Preview query results before export
- 以 UI 建立篩選條件，不需寫 SQL
- 支援全域 AND / OR 條件邏輯
- 畫面清楚顯示目前條件邏輯
- 匯出前可先預覽符合條件的資料筆數
- Reduce dataset size by filtering rows
- Export results as CSV or XLSX
- Export progress indicator to avoid “is it frozen?” confusion
- 透過篩選條件縮小資料集
- 支援 CSV / XLSX 匯出
- 匯出時提供進度提示，不再誤以為系統當機
- Basic exploratory analysis and summaries
- Designed for practical data preparation, not heavy BI workloads
- 提供基礎探索式分析與摘要
- 著重資料前處理，而非取代完整 BI 平台
-
Backend
- Python
- FastAPI
- DuckDB
-
Frontend
- Vanilla HTML / CSS / JavaScript
- No heavy frontend framework
-
Storage
- Local filesystem
- Input:
data/input/ - Output:
data/output/
dataset-workbench/
├── src/
│ ├── api/ # FastAPI routes
│ ├── pipelines/ # Data processing & filtering logic
│ └── frontend/ # UI (HTML / CSS / JS)
├── data/
│ ├── input/ # Place input datasets here
│ └── output/ # Exported results
├── README.md
└── .gitignore
- Python 3.9+
- pip
pip install -r requirements.txtuvicorn src.api.main:app --reload- Analysis UI（分析工具）: http://localhost:8000/ui
- Dataset Reducer（CSV / XLS 縮小器）: http://localhost:8000/reduce
- Upload a CSV / XLS file to the UI.
- Preview schema and sample rows without loading everything into Excel.
- Build filter conditions using AND / OR logic.
- Preview how many rows will be matched.
- Export a reduced dataset for downstream analysis.
流程說明：
- 上傳 CSV / XLS 檔案
- 快速預覽欄位與資料樣本
- 使用 AND / OR 條件建立篩選邏輯
- 匯出前先確認符合條件的資料筆數
- 匯出縮小後的資料集供後續分析使用
-
data/input/- Place source CSV / XLS files here.
- 原始資料請放置於此目錄。
-
data/output/- Exported datasets will be written here.
- 匯出後的結果檔案會產生於此目錄。
Note:
- The directory structure is tracked by Git.
- Actual data files are intentionally ignored to avoid committing sensitive data.
注意：
- Git 只追蹤目錄結構
- 實際資料檔案不會被提交到版本庫
- This project is not a full BI or visualization platform.
- Filter logic currently supports global AND / OR only (no nested groups).
- Designed for interactive use and data preparation, not long-running batch jobs.
限制說明：
- 本專案並非完整 BI 或視覺化平台
- 目前僅支援全域 AND / OR 條件，尚未支援巢狀群組
- 著重互動式資料整理，不適合長時間批次處理
This project is under active development. Features are added based on real-world usage rather than theoretical completeness.
本專案持續開發中，功能演進以實際使用情境為優先考量。