Setting up your PC for machine learning without anaconda.
Anaconda is a really handy package that include most of the needed libraries for Machine learning and comes with jupyter notebook.
When I started to set up my work windows machine I found that due to the older windows version and some pre dependencies I am having problems with installing packages with -conda command.
After scouring stack overflow I was getting a lot of this quick fixes like changing the path or upgrading command, or not using conda command at all. And I found that my machine has already a lot of libraries pre installed. So basically anaconda is adding a lot of overhead or worse various existing versions and paths are becoming complex for machine to handle.
If you are getting fed up with the installation and want to get to the real work quick this is for you.
First check if python is installed on your device and the version. I would recommend having atleast 3.5+ version and set the path properly. You can find a lot of online tutorials for this. As all systems usually have python in this tutorial I am going to skip it.
Check if pip is installed or not by using this command in terminal
pip help or pip3 help
If you get a result like command not recognised. install it using this command.
python get-pip.py
If you have pip installed upgrade pip
pip install --upgrade pip
After pip is properly installed open an editor and import the libraries you need for your machine learning projects
But before that if you are going to use Pytorch or tensorflow I will suggest to install it now by following the official documentation. They also use pip. No worries!
Now check which of these are not modules are not installed
import numpy
import pandas
import matplotlib
import sklearn //Scikit-learn
import cv //opencv //image processing library
I found that most of the basic libraries were already there after python and pip update. Check which of these are giving module not found error and install them accordingly.
pip install numpy
pip install pandas
pip install opencv-python
pip install matplotlib
pip install scikit-learn
pip install scikit-image
Keep in mind ML has a lot of libraries but good news is whenever you are working with them you need to import them first and if there are module not found error you can promptly install the library.
Next for editor you can install vscode/ sublime text or jupyter notebook. You don’t need to be inside any separate environment to work on ML projects and it is going to save a bit of space on your machine.