NumPy is both one of the most important and confusing python modules. It provides advanced mathematical capabilities which can run in a fast and efficient manner. In fact, it’s so efficient that the module’s something of a standard within python libraries. If a python library offers any kind of advanced functionality then there’s a very good chance that it makes use of NumPy. This is so common that most people’s first introduction to NumPy usually comes from its use by another module. And this is also why someone’s first introduction to NumPy will usually come from an error message reading modulenotfounderror: no module named numpy.
What is causing this error?
The error message tells us that the python runtime is looking for a module named NumPy but can’t find it. We often encounter this error when using a third-party library that tries to import the NumPy module or a similar python package. For example, consider the case of someone writing a simple python program that uses NLTK.
Anything related to artificial intelligence will make heavy use of advanced math, and python’s natural language toolkit is no exception to that rule. As such it requires the python package file NumPy. If we don’t have the NumPy file installed, but import NLTK, then we’ll receive the module not found error. The same goes for any other module which requires any NumPy version.
Solution
Resolving this error is as simple as installing the correct NumPy version for your python environment variable. Different operating systems typically provide their own means of installing the module. However, it’s usually best to install NumPy by using pip. In theory, this means that we simply need to type “pip install numpy” onto the command prompt. After all, the newer version pip package installer was created in large part to get around the issues related to multiplatform design, and avoid the type of import error problem we have here. There are still a few caveats we need to consider when using pip to install NumPy.
If we were installing NumPy through an operating system’s typical package manager system we’d have to worry about the version of NumPy that will be installed, due to the multiple python version scenario. Pip has the answer to that problem. However, keep in mind that it’s usually best to avoid mixing and matching installation sources.
If you typically install python packages through your operating system’s package manager command prompt rather than the python prompt or a different package/project interpreter, then you might want to continue doing so with the NumPy package. This is particularly true for Linux users due to python’s root user ubiquity within that platform. For example, using apt to install a python program that uses NumPy might automatically install it as a dependency. If NumPy had already been installed through pip then we might end up with problems relating to versioning or installation failures, such as this np importerror. Linux distributions also tend to provide multiple versions of pip, and may have a multiple python version issue as well. As such, you might need to launch it by typing pip3 to ensure you’re installing the NumPy package for your python3 runtime rather than python’s 2.x branch. For example, you might need to type ‘pip3 install numpy’ to get ther correct version for your python environment.
Windows has a potential stumbling block as well. When you’re writing python code on windows you should make sure your python path has been set properly. Go into ‘This PC’ and select ‘Properties. Now select ‘Advanced system settings. Proceed by selecting ‘Environment variables’. Select ‘Path’ under ‘System variables. You’ll need to make sure that your python installation directory is listed within Path. If it isn’t then your system might not look for NumPy in the correct location.
After you’ve gone through these steps you can test the results. Just load up python on the command line. Then type import numpy. This will load up the NumPy library. It should load up with no errors and return you to the python interpreter.