我打造 FixMyCSV：以 AI 將零售業雜亂資料整理成可靠數據
I Built FixMyCSV to Turn Messy Retail Data into Trustworthy Data with AI
開發者打造 FixMyCSV，這款命令列助手使用 Google Gemma，協助小型零售商檢查及清理 CSV 資料。它可識別缺失值、格式不一致、重複記錄及負數，並讓用戶預覽和核準支援的修改後再儲存清理副本。清理流程結合 AI 建議與按固定規則執行的 Python 操作，並以 Backboard 儲存業務專屬規則、用 SQLite 保留操作記錄。
This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle
What I Built
FixMyCSV is a command-line data cleaning assistant designed for small retail businesses that manage inventory and other business data in CSV files.
It helps users identify data quality issues such as missing values, inconsistent formatting, duplicate records, and negative numeric values. It uses Google Gemma to suggest possible cleaning actions and allows users to preview and approve supported changes before saving a cleaned copy.
FixMyCSV also supports storing business-specific cleaning rules through Backboard and maintaining a history of cleaning operations using SQLite.
The goal is to help small retailers improve data quality without requiring advanced technical knowledge, while keeping them in control of changes to their data.
Demo
Code
How I Built It
I built FixMyCSV using Python and Pandas for CSV inspection, data-quality checks, and cleaning operations. I used Google Gemma to generate AI-assisted cleaning suggestions based on the detected issues, with the model accessed through the Google GenAI API.
I integrated Backboard to store and retrieve business-specific cleaning rules, allowing relevant rules to provide context for future suggestions. The application uses Rich and Questionary for its interactive command-line interface and SQLite to maintain cleaning history.
The cleaning process combines AI-generated recommendations with deterministic Python operations. Users can review proposed changes and approve them before a cleaned copy is saved, keeping the original data safe.
Why Does Open Innovation Matter?
Open innovation made it possible to build FixMyCSV by combining open-source tools and AI with a practical data-cleaning workflow. Using Gemma gave the project access to a model for generating cleaning suggestions, while Python and Pandas handled data processing deterministically.
This approach makes the project easier to inspect, modify, and extend. Developers can adapt the cleaning logic, add support for new data-quality issues, and improve the workflow without rebuilding everything from scratch. It also leaves room to explore different models as requirements change.
For small retailers, this means a more accessible way to improve data quality, with AI assistance and human approval working together rather than relying on automatic, unchecked changes.
Prize Categories
- Best Use of Gemma
- Best Use of Backboard
來源：DEV Community · dev.to