CSV Data Profiler
Understand a new CSV in seconds: types, gaps, ranges and common values.
How to use the CSV Data Profiler
- Drop a CSV file onto the panel or click to choose one.
- Every column gets a card showing its type, how full it is and its key figures.
- Use the filter box to find a column in wide files.
- Download the summary as a CSV or the full profile as JSON.
About this tool
Before you clean, chart or import a dataset, you need to know what is in it. Opening a big export in a spreadsheet tells you little and may even crash the program. This profiler reads the file once and builds a summary card for every column: what kind of data it holds, how many cells are empty, how many different values appear, the lowest, highest and average numbers or the earliest and latest dates, the most common values and a small chart of how values are spread.
It is built for large files. The file is streamed in pieces and every statistic is a running total, so even a file of several hundred megabytes needs only a little memory. Counts stay exact for typical columns; for columns with millions of different values, the tool switches to well-known estimation methods and marks those figures as approximate.
Watch for columns labelled Mixed. They usually hide a problem, such as a number column with a few "N/A" entries or dates typed in two formats, which will cause errors when the data is imported into a database or analysis tool.
Frequently asked questions
How big a file can it handle?
Files of hundreds of megabytes work on a normal computer, because the file is read in streaming chunks instead of all at once. The limit is mostly how long you are willing to wait.
How does it decide a column's type?
Every filled cell is checked to see whether it reads as a whole number, a decimal, a date or a yes/no value. Only when every filled cell agrees does the column get that type. If even one cell differs, the column is treated as text and labelled Mixed, with a count of each kind.
Why are some counts marked "about"?
To keep memory use small, exact counting stops once a column has a very large number of different values. Beyond that point the tool estimates, typically to within a few percent.
Does it work with semicolon or tab separated files?
Yes. The delimiter is detected automatically, and you can set it yourself. Files using a semicolon delimiter are assumed to use a comma as the decimal mark, as is common in Europe.
Is my data uploaded for analysis?
No. The file is read by your browser and the statistics are calculated on your device, which makes the profiler suitable for customer records and other confidential data.