Data Binning Tool

Bin a numeric column of a CSV into buckets

Paste a CSV, pick one numeric column, and this tool sorts each row into a bucket and labels it — locally in your browser, nothing is uploaded. Choose how the buckets are drawn:

• Equal-width — split the column's value range into buckets of the same width. bins sets how many (default 4 = quartiles). Simple to read, but a few extreme values can leave some buckets nearly empty.
• Quantile — draw the edges so each bucket holds roughly the same number of rows (equal-frequency). Preferred for skewed data. Duplicate edges from repeated values are merged automatically, so you may get fewer buckets than requested.
• Custom edges — supply your own strictly-ascending boundaries (like 0,18,65,120). Values below the first or above the last edge get a blank label.

Label each bucket yourself (Custom bucket labels, one per bucket) or let the tool auto-label them as an interval range (like (50, 75]) or a 1-based bucket index. Right-closed intervals decides whether a boundary value falls in the lower or upper bucket, and Output mode either appends a new <column>_bin column or replaces the source column with the label.

Worked example

With the default Equal-width method and 4 buckets on the score column, this input:

name,score
Ann,0
Bo,40
Cy,70
Di,100

becomes:

name,score,score_bin
Ann,0,"[0, 25]"
Bo,40,"(25, 50]"
Cy,70,"(50, 75]"
Di,100,"(75, 100]"

score runs 0–100, so four equal-width buckets have edges 0, 25, 50, 75, 100. 40 lands in (25, 50] and 70 in (50, 75]. The first bucket is written [0, 25] because the lowest edge is always included; interval labels that contain a comma are quoted by the CSV writer.

Limits & edge cases

• The chosen column must be numeric — every present value has to parse as a finite number, or the run errors. NaN and inf count as non-numeric.
• Blank cells stay blank: an empty target cell gets no label and is left out of the edge statistics.
• With Custom edges, a value outside the first/last edge gets a blank label rather than being forced into an end bucket.
• Quantile binning can merge duplicate edges on skewed or repeated data, so the result may have fewer buckets than bins.
• A constant column (all values equal) yields a single bucket.
• Custom bucket labels must supply exactly one label per bucket, or the run errors — handy because quantile merging can change the bucket count.
• Turn off First row is a header for files with no header row; then the column selector must use a 1-based index and row 1 is binned like any other.
• Everything runs in memory, so very large files are bounded by your browser's available memory.

FAQ