Store the data
Keep the CSV you downloaded in a folder on your laptop. Leave that original file unchanged. Do your edits in a second copy. Save a note beside the file with the source URL and the download date.
Pick a home for your data
These sizes are comfortable starting points. A very wide table, or an older laptop, may need a smaller file. The sizes below are for files on your laptop. A free chat may accept only a smaller upload. GitHub is a website that stores files and their history. A repository is a folder that remembers every change. Parquet is a compact file format for big tables. SQL is a language for questions about a table.
| Method | Good for | Size it handles comfortably | Easy to share | Free | Needs code |
|---|---|---|---|---|---|
| CSV file | One small table you can upload to a chat | Small tables, up to a few dozen megabytes | Yes, send a copy | Yes | No |
| Google Sheets | Classmates editing one table together | Small tables, up to tens of thousands of rows | Yes, share a link | Yes, with an account | No |
| SQLite or DuckDB | Questions across larger tables | Hundreds of megabytes to a few gigabytes, depending on your laptop | Yes, share the database file | Yes | Usually SQL |
| Parquet | Storing a large table in a smaller file | Large files, hundreds of megabytes to many gigabytes, with a suitable tool | Yes, share the file. The reader needs a tool that opens Parquet | Yes | Usually |
| GitHub repository | A public history of small files and code | Small CSVs, a few megabytes each | Yes, a public link when the licence allows | Yes, with an account | No, for uploading the file |
Three questions to choose
- How big is it?
A small table can stay a CSV. If the file is too large to open, use SQLite or DuckDB to ask questions, or Parquet to store it in less space.
- Who needs it?
Keep it on your laptop if it is only for you. Use Google Sheets if classmates will edit it together. Use a GitHub repository for a small public file when the licence allows sharing.
- Will it change?
A one-time download needs a dated copy. Each later download needs its own raw file and source note. Use GitHub for the history of small text files. Use a database when the table keeps growing.
Choose a storage method
✅ Free pending testPaste this prompt into your AI chat, and replace each [bracket].
My public dataset is [file size and approximate number of rows]. It will be used by [just me / classmates / the public] and updated [once / monthly / often]. Recommend one option: CSV, Google Sheets, SQLite or DuckDB, Parquet, or a GitHub repository. I am a beginner using free tools. Explain your choice in three short sentences and say whether it needs code. Do not assume I may redistribute the data.The reply names one storage option and gives a reason about size, who uses the file, and how often it is updated.
Make a working copy
✅ Free pending testDuplicate the downloaded file and add the word working to the copy name.
My downloaded file is called [your filename]. Suggest a clear name for a working copy that keeps the same file extension. The original download must stay unchanged.You see two files in the folder: the original download, and a copy with working in its name.
Save a source note beside the download
✅ Free pending testSave this note as source.txt in the same folder, and replace each [fill in] that you know.
Give me a short plain-text source note to save beside my raw download. Include filename, publisher, dataset and edition, source URL, download date in YYYY-MM-DD form, selected countries and years, units, licence, and row count. Leave unknown values as [fill in]. Do not invent them.You see source.txt in the same folder as your download, and the note shows the source URL, the download date, and [fill in] where a value is still blank.
Write the download date as year-month-day, for example 2026-10-10.
Next: explore inflation.