Convert HDF5 to CSV

HDF5 (.h5 or .hdf5 — same format, different extension) isn’t a table, it’s a container: datasets organized into groups, like folders within a single file. That means “convert to CSV” really means “pick a dataset and export it” — one HDF5 file can yield several CSVs. There’s no in-browser converter for this format here yet; the free two-step route below works for any HDF5 file. Python users can jump straight to that section.

No installation: find the dataset, then export it

Step 1: Find your dataset path with myHDF5

Go to myhdf5.hdfgroup.org — the HDF Group’s free in-browser viewer. It runs entirely client-side (your file is never uploaded), so it’s safe even for confidential data. Open your file, browse the tree, and note the path of the dataset you want, e.g. results/temperature.

Step 2: Export it with Google Colab

1. Open Colab — Go to colab.research.google.com, sign in with any Google account, and click New Notebook.
2. Upload your file — Click the folder icon in the left sidebar, then the upload button, and select your .h5 file.
3. Run the code — Paste the snippet below into the cell and press the run button.
4. Download the result — Refresh the file sidebar, right-click output.csv, and choose Download.

import h5py
import pandas as pd

with h5py.File("yourfile.h5", "r") as f:
    f.visit(print)  # lists every dataset path — handy if you skipped step 1
    data = f["results/temperature"][()]  # replace with your dataset path

pd.DataFrame(data).to_csv("output.csv", index=False)

Unlike myHDF5, Colab does put the file on Google’s servers (in a temporary private session that disappears when the tab closes) — for confidential data use HDFView below instead.

Free desktop app: HDFView

HDFView is the HDF Group’s official free desktop browser (Windows/macOS/Linux, from hdfgroup.org). Open the file, double-click a dataset to view it as a table, then use Table → Export Data to Text File. Note the output is a delimited text file rather than a ready-made .csv — fine for a quick look, but the Colab route gives you a cleaner CSV.

Python

Two cases, and mixing them up is the usual failure:

• The file was written by pandas (df.to_hdf(...)): read it back with pandas.read_hdf().
• Any other HDF5 file (instruments, simulations, other languages): read_hdf will fail — use h5py, list the dataset paths, and pick one.

import pandas as pd

df = pd.read_hdf("data.h5")  # pandas-written files only
df.to_csv("output.csv", index=False)

For generic files, use the h5py snippet from the Colab section — it’s identical locally (pip install h5py pandas first).

R

rhdf5 exists on Bioconductor, but unless you’re already in that ecosystem the install friction isn’t worth it — use the Colab route instead.

Which method should I use?