What it does
Before you analyse a data set, you need to know what is in it: which columns are numbers, how many cells are empty, whether a typing slip has crept in, which columns move together. This profiler answers those questions for a CSV or Excel file in a few seconds, in your browser. The file is never uploaded.
For the whole table it shows the number of rows and columns, missing cells, rows with a gap and exact duplicate rows, plus a list of things to check. For every column it finds the type (number, date, yes/no or text) and gives:
- Present, missing and unique counts.
- For numbers: mean, standard deviation, minimum, quartiles, median, maximum, skewness, excess kurtosis, how many zeros and negative values, and possible outliers (values more than 1.5 times the interquartile range beyond the quartiles), with a histogram.
- For text, dates and yes/no: the most common values with their counts and shares, and the first and last value.
A correlation heatmap shows how every pair of number columns moves together, with the strongest pairs listed.
How to use it
- Open a CSV, TSV, text or Excel (.xlsx) file, or paste a table. The first row should hold the column names; if it is all numbers, the columns are called Column 1, Column 2 and so on.
- Read the overview and the Things to check list first.
- Pick a column under One column in detail to see its histogram or its most common values.
- Choose Pearson correlation for straight-line relationships, or Spearman for any steadily rising or falling trend. Spearman works on ranks, so one wild value cannot dominate it.
- Download the column statistics as CSV, or the whole report as a web page (HTML), image, PDF or Word file.
Histogram bins are set automatically with the same rule as numpy (the narrower of the Sturges and Freedman-Diaconis widths), or you can fix 10, 20 or 30.
Worked example
The example is a made-up table of 41 plants. The page shows 41 rows × 8 columns, 5 missing cells and 1 duplicate row (plant P11 appears twice). The column sown is found to be a date, fertilised a yes/no column, and region text with 1 gap and East as its most common value (12 of 40).
height_cm has 2 missing values (one empty, one written NA). Its mean is 43.67 cm with an SD of 13.49, but the median is 42 cm: the list flags 2 possible outliers, and the histogram shows one bar far out at 118 cm, a likely slip for 11.8 or 48.0. That one value also pulls the Pearson correlation between height and mass down to 0.540; switch to Spearman and it is 0.845. Fixing or removing the slip is the first job before any analysis.
Good to know
- Statistics follow pandas: the standard deviation divides by n − 1, quartiles interpolate linearly, and correlations use the rows where both values are present.
- A column with numbers and a few words (such as “n/d”) is treated as text, and the list says how many cells to fix.
- Dates are recognised when written year first, such as 2026-03-01 or 2026/3/1 14:30. Other date styles are treated as text, because 03/04 could be March or April.
- Files up to 15 MB, 60 columns and 50,000 rows. The heatmap shows up to 25 number columns.
What it does not do
It does not change your data, guess at mixed date formats, or test whether a correlation is significant. Clean the data in your spreadsheet or with SQL, then run the profile again.
Frequently asked questions
Is my file uploaded? No. It is read by your browser and nothing is stored or sent anywhere.
Why is a number column called text? At least one cell is not a number. The Things to check list shows an example cell.
What do skewness and kurtosis mean? Skewness above 0 means a long tail to the right; excess kurtosis above 0 means heavier tails than a normal distribution.
Where the numbers come from
- Statistics as defined in pandas (describe, skew, kurt, value_counts, duplicated, DataFrame.corr) and numpy's automatic histogram bins; checked against pandas 3.1 and numpy 2.5 on the example and a random table.
- The example table of 41 plants is made up for this tool.
Standard scientific values, such as constants, atomic weights and melting points, are facts found in every textbook; we name where ours were checked. If you are an author or publisher and want a value, a credit or a table changed or removed, tell us and we will act quickly.
Last updated 2026-11-09.