CSV Statistics Calculator

Calculate descriptive statistics for one numeric CSV column with explicit missing-value and variance conventions.

Your CSV stays in this browser. No input is uploaded.

Comma-separated CSV: up to 200,000 characters, 200 unique nonempty headers and 10,000 data rows.

Decimal numbers only; no currency, grouping commas or percentages. Skipping is explicit. Floating-point results can lose precision; sample variance is undefined with one retained value.

Summarize one numeric CSV column

Paste a CSV table, select a Numeric column, and choose Calculate statistics. The result reports count, sum, mean, median, minimum, maximum, variance and standard deviation. It also reports blank and unsupported cells separately. Results can be copied or downloaded as JSON for inspection.

Example: compare population and sample variance

Load the example and choose score. Its values are 2, 4, 6 and 8. Count is 4, sum is 20, and both mean and median are 5. With the default population convention, variance is 5 and standard deviation is approximately 2.236. Enable Use sample variance (n − 1) to obtain variance 20/3, approximately 6.667, and standard deviation approximately 2.582.

Make missing-value decisions explicit

By default, a blank or unsupported cell causes an error. Skip blank and nonnumeric cells includes only supported numeric values and displays how many were excluded. This is not imputation: missing values are not treated as zero, and skipped text is not silently coerced. The retained numeric count, rather than all input rows, is the denominator. An all-excluded selection cannot produce a report.

Numbers may have surrounding whitespace, a sign, decimal point or exponent. Currency symbols, grouping commas, percentages, hexadecimal notation, Infinity and NaN are unsupported. Values use JavaScript floating-point arithmetic; very large integers and high-precision decimals can lose precision. An overflowing calculation is rejected. These are descriptive summaries, not a recommendation or a test of statistical significance.

Very small magnitudes can also lose information. A nonzero token such as 1e-400 that would parse as zero is unsupported; it causes an error unless skipping is enabled, in which case it is counted as unsupported. If distinct parsed numbers produce zero variance through underflow, the calculation stops with a precision error. Rescale values into a practical unit before calculating; variance then changes by the square of the scale factor. Other floating-point rounding can still make distinct original decimal strings equal, so a zero variance does not prove the source measurements were identical.

Variance conventions

Population variance divides squared deviations by n. Sample variance divides by n − 1. With one retained number, population variance is zero and sample variance and standard deviation are undefined. Choose the convention appropriate to what your dataset represents; the checkbox does not determine whether your data was collected through valid sampling.

Input limits are 200,000 characters, 200 unique nonempty headers and 10,000 data rows. CSV is parsed locally and is not uploaded. Editing input or options clears old results. Use CSV Column Type Inspector to review mixed fields first, or CSV Row Filter to select the records you intend to summarize.