Histogram maker from CSV

The automatic rule gives 34 narrow bins for 8,760 values. Both show the same broad shape, highest around 12 °C with a long tail of cold winter hours; 20 wider bins smooth it, and the most common range moves with the edges.

The same in Python

With the sample file downloaded, these lines give NumPy’s answer for the same bins:

import numpy as np, pandas as pd

t = pd.read_csv("megarows-sample-leeds-hourly-weather.csv", skiprows=[1])["Leeds_Temperature"].to_numpy()
for bins in ["auto", 20]:
    counts, edges = np.histogram(t, bins=bins)
    k = counts.argmax()
    print(f"{bins}: {len(counts)} bins of {edges[1] - edges[0]:.4f} °C, "
          f"most common {edges[k]:.2f} to {edges[k + 1]:.2f} °C ({counts[k]} hours)")

They print:

auto: 34 bins of 0.7415 °C, most common 11.50 to 12.24 °C (502 hours)
20: 20 bins of 1.2605 °C, most common 11.80 to 13.06 °C (814 hours)

We ran them on 9 October 2026 with numpy 2.1.3, pandas 2.2.3, Python 3.12.15. A test works out the histogram on every change to the site and checks that every edge and count is NumPy’s, and that the panel’s numbers are the ones on this page. Download the sample file (CSV, 299 KB).

How the bins are chosen

Every bin has the same width, and together they run from the smallest value of the column to the largest. A value on the edge between two bins counts in the upper one; the largest value counts in the last bin. These are NumPy’s rules (numpy.histogram), so the counts match a histogram made in Python.

• Automatic (the default) uses NumPy’s “auto” rule: it works out the Freedman–Diaconis width (twice the interquartile range divided by the cube root of the count, robust to outliers) and the Sturges width (the range divided by log₂ n + 1, good for small samples), and takes the narrower. At most 500 bins are drawn (a rule that asks for more gets 500), and if a few extreme values make it ask for more than ten million, the histogram uses 100 bins instead.
• 10, 20, 50, 100 or 200 bins, from Plot settings, Axes, Bins, split the same range into that many equal parts.

For Freedman–Diaconis, Sturges or square-root bins on their own, use Bins by a rule over the plot: the Math dialog’s Histogram gives the bins as a table with the rule you pick.

Counts, the table and the statistics

The height of each bar is the number of values in that bin. Show as a table in the plot’s menu lists every bin with its From, To and Count, and its Download CSV saves exactly those rows. The statistics panel shows the values counted, the number of bins, the most common range and, when there are any, the rows with no value, which are left out.

The histogram always covers the whole file, whatever the zoom. It shows counts, not percentages or a density: to compare two files of different lengths, divide by the values counted.

Checked against NumPy

Private by design: nothing is uploaded

This tool reads your file and counts the values in your browser, so the file and its name never leave your device. Files with millions of rows work: every value is counted, and the practical limit is your device’s memory.