Plot multiple CSV files on one graph

Three test runs, a week of daily logs, readings before and after a change: comparing them usually means one CSV file per run, each with its own time column, column names and layout. This guide shows how to put several CSV files on one chart in your browser, one series per file and each labelled with its file, with three example runs to try. It also covers when to overlay, merge or join, how to check columns and units, what happens when one file has dates and another elapsed seconds, and how gaps, sample rates and large files are drawn.

The example is three made-up 20-minute runs of weather masts at Leeds, York and Hull, written the way different loggers write: the Leeds file has dates and times (UTC), from 08:00:00, one row a second; the York file seconds from its start, with no date, ten rows a second, with a 30-second pause; the Hull file dates and times (UTC), from 08:00:30, one row every 2 seconds.

Overlay, merge, join or compare differences?

“Putting CSV files together” can mean four different jobs. Pick the one that matches your files before you start:

The merge tool appends files in your browser when the parts belong to one log. For separate runs, keep one series per file: merging them would join the end of one run to the start of the next.

Check each file’s columns and units

Files from different loggers rarely agree on names. In the example the same temperature is Leeds_Temperature_C in one file, temperature_cin another and Temperature (C) in the third, and the columns come in a different order. The plotter matches columns across files only by the same name, in any case, so here you tick each file’s temperature yourself. When the names do match, the files open with that column of each already on the chart, and a column’s menu (a right-click on it in the list) offers “Plot from every file”.

The first lines of each file:

leeds_mast_run1.csv
Time,Leeds_Temperature_C,Leeds_Wind_Speed_ms,Leeds_Pressure_hPa
2026-03-02T08:00:00.000Z,3.4,4.9,1008.0

york_mast_run2.csv
time_s,wind_speed_ms,temperature_c,pressure_hpa
0.0,7.0,3.5,1008.4

hull_mast_run3.csv
# Hull mast, synthetic
Timestamp;Temperature (C);Wind speed (m/s)
2026-03-02T08:00:30.000Z;3,7;6,1

The Hull file is written the European way: a comment line above the header, semicolons between the columns and decimal commas. The plotter reads it as it is: the line starting with # is skipped, the semicolons are found, and 3,7 is read as 3.7, so there is nothing to convert first. It opens with 600 rows and 3 columns.

Check the units too. All three runs here are in °C, m/s and hPa, with the unit in the column’s name. A chart can’t tell that one file holds °F and another °C, so convert one of them first, in the file or with a formula in Math.

Absolute time or elapsed time for each run

Leeds and Hull write dates and times; York writes seconds from its own start, in a column called time_s, with no date. How they share an axis depends on that:

• Dated files keep their clock. Leeds and Hull on one chart are drawn by their timestamps, 08:00 to 08:20 on 2 March 2026, and Hull’s line starts 30 seconds after Leeds’s, as it did.
• Relative lines the runs up at 0. Press Relative next to Link X in the toolbar (“Relative | File”): each file’s times are counted from its own first timestamp, the axis reads in elapsed hours, minutes and seconds, and the chart’s corner says where 0 is.
• A file of elapsed seconds has no clock. With York in the set the files’ times are of different kinds, so each run is drawn from its own start, in elapsed time, and the corner says “3 files · Aligned from each file’s start: the files’ times are of different kinds”. Hull’s late start is not shown then; compare the dated files on their own when the clock matters.

A plot of several files has the same choice in Plot settings → Axes → X alignment: Automatic, Same X values, From each file’s start or Row number. Nothing is shifted or stretched to make runs fit: each reading keeps its own time.

Keep each run identifiable

A chart of several runs is only useful while you can tell them apart. Each file keeps one colour, the colour of its tag on its chip, and each series of that file takes it where it is free, so a run keeps its colour from chart to chart. The file’s name follows its series into the chips, the statistics, the Compare tables and the legend of a saved PNG.

• Name files by site and run before you open them, as leeds_mast_run1.csv does, rather than export (3).csv.
• Rename a file in the plotter from its chip’s ⋯ menu → Rename, for example to “Leeds run 1”: the legend and the image use the new name, and the file on your disk stays as it is.

Gaps, sample rates and large files

The runs need not share a rate. Leeds logs once a second, York ten times a second and Hull every 2 seconds, and each reading is drawn at its own time: nothing is resampled or interpolated, and a slower run has fewer points.

York stopped logging for 30 seconds, from 584.9 to 615.0 seconds. The pause keeps its real length on the time axis, and the line joins the last reading before it to the first after it with one straight stroke, plain to see in the noisy trace; the data table shows the jump. Only an empty cell leaves a break in a line, not a missing row. To see every reading as a dot, so the pause shows as an empty stretch, switch the plot to a scatter chart in its ⋯ menu.

Each file is read on your device, and a file of millions of rows opens like a small one: the chart shows an overview that keeps every peak and fills in full detail as you zoom. Up to eight files can be open and up to eight series can share a chart, and the plotter asks first when the files together may not fit in the device’s memory. Our guide shows how to plot a large CSV file.

Compare, then the next step

With the runs on one chart, “Compare across files” in the statistics panel, or the Compare tab under the plots, puts every statistic of each file side by side with the differences worked out, and lines the rows up at the same X. Each table downloads as a CSV. To read through every row of a run first, open it in the large CSV viewer, which shows the data table first.

For a chart you rebuild every week, or dozens of runs, a script repeats the same steps every time. In Excel, put each run on a sheet, make an XY (scatter) chart of the first and add the others with Select Data, as the FAQ below explains. In Python, pandas reads each file with its own settings and Matplotlib draws them on one axes, each from its own start:

import pandas as pd
import matplotlib.pyplot as plt

def elapsed(times):
    return (times - times.iloc[0]).dt.total_seconds()

leeds = pd.read_csv("leeds_mast_run1.csv", parse_dates=["Time"])
york = pd.read_csv("york_mast_run2.csv")
hull = pd.read_csv("hull_mast_run3.csv", sep=";", decimal=",", comment="#", parse_dates=["Timestamp"])

fig, ax = plt.subplots()
ax.plot(elapsed(leeds["Time"]), leeds["Leeds_Temperature_C"], label="leeds_mast_run1")
ax.plot(york["time_s"], york["temperature_c"], label="york_mast_run2")
ax.plot(elapsed(hull["Timestamp"]), hull["Temperature (C)"], label="hull_mast_run3")
ax.set_xlabel("Seconds from each run's start")
ax.set_ylabel("Temperature (°C)")
ax.legend()
fig.savefig("three-runs.png")

These lines ran on the three example files with pandas 3.1 and Matplotlib 3.11 (Python 3.12). Matplotlib draws a straight stroke across York’s pause as well.