When you receive a CSV you usually want a quick sense of each column before doing anything with it: what range the numbers span, where the average sits, and how many cells are empty. This CSV column statistics tool parses your file and produces per-column summaries in your browser, with a small histogram for the numeric columns.
How it works
The tool first parses the CSV with a quote-aware RFC 4180 parser, so quoted fields that contain commas, newlines, or doubled quotes are handled correctly instead of naively splitting on commas. You choose the delimiter and whether the first row is a header.
For each column it counts non-empty and empty cells. If every non-empty cell is numeric it computes:
- minimum and maximum,
- mean (arithmetic average),
- median (middle value, or the average of the two middle values),
- population standard deviation, the square root of the average squared deviation from the mean,
- a ten-bin histogram rendered as a sparkline.
Columns with any non-numeric value are summarised instead by their count of distinct values and a few sample entries.
Example
In the sample, the score column has one empty cell, so its null count is 1, and its statistics are computed over the remaining numeric values. The city column is text, so it reports the number of distinct cities and a list of samples rather than a mean.
Reading the output
Each column produces a card. For numeric columns, the card shows:
- Min / Max — the range. A very wide range can indicate outliers worth investigating.
- Mean — the arithmetic average. Compare it to the median: if they diverge significantly, the distribution is skewed (for example, a salary column with a few very high earners pulls the mean above the median).
- Median — the middle value; robust to outliers and often more useful than the mean for skewed data.
- Std dev — how tightly values cluster around the mean. A low std dev means values are similar; a high one means they spread widely. Reported as the population standard deviation; apply Bessel’s correction (
multiply by sqrt(N / (N-1))) if you need the sample value. - Null count — empty or missing cells in that column. High null counts flag data quality issues.
- Histogram — a ten-bucket sparkline showing the shape of the distribution. A left-skewed peak with a long right tail is a typical pattern for income, page load times, and similar real-world data.
For text columns, the card shows the number of distinct values and a sample, which is useful for spotting inconsistent encodings (“UK”, “United Kingdom”, “uk”) before a group-by.
When to use this
Run the column stats tool at the start of any data cleaning or analysis project:
- Before importing into a database — spot null columns and confirm numeric types are actually numeric.
- Before training an ML model — identify skewed distributions, outlier-heavy columns, and missing data that need preprocessing.
- After a data pipeline run — quickly confirm that aggregation counts match expectations and no column unexpectedly went to all-nulls.
- In a spreadsheet audit — faster than building formulas by hand for each column.
Everything runs locally, so even confidential CSVs stay on your device.