Jetson Nano for Data Science

Has anyone been able to install Numpy, Pandas, Scipy, Scikit-Learn and Keras (besides Tensorflow) in order to do data science? It’s UNBELIEVABLY HARD to do so. I’m way past a few dozen attempts but eventually something always fails along the way. If so, please, share some tips from start to bottom.

NVidia really should do something about this: why do you even release a special Ubuntu distro that can barely hold itself together?


It takes time to install these package since most of them need to build from source.

1. Numpy

$ sudo pip3 install numpy

2. Pandas

sudo pip3 uninstall enum34
sudo pip3 install pandas

3. Scipy

tar -xzvf scipy-1.3.3.tar.gz scipy-1.3.3
cd scipy-1.3.3/
python3 install --user

4. Scikit-Learn

sudo apt-get update
sudo apt-get install -y build-essential libatlas-base-dev gfortran
sudo apt-get install python3-sklearn

5. Keras
Install TensorFlow with the steps here.

$ sudo pip3 install keras


Hi! To further ilustrate my point, I’m trying to install Numpy (the first package of all) and I get this error:

Command “/usr/bin/python3 -u -c “import setuptools, tokenize;file=’/tmp/pip-build-kzxf68kb/numpy/’;f=getattr(tokenize, ‘open’, open)(file);’\r\n’, ‘\n’);f.close();exec(compile(code, file, ‘exec’))” install --record /tmp/pip-s4_6g9yo-record/install-record.txt --single-version-externally-managed --compile” failed with error code 1 in /tmp/pip-build-kzxf68kb/numpy/

Any clues?

Update: nevermind, installed Cython and it worked now. I’ll keep on installing the packages, hopefully I can make everything work as it should.


Not sure which JetPack do you use.
But the issue in comment#3 should be fixed by updating the setuptools.

sudo pip3 install --upgrade setuptools