My python data analysis environment is as follows.
os fedora22。 I couldn't stop the installation in the middle of the method of burning to DVD, so I upgraded from fedora21. fedora22 is the best.
python It is anaconda 64bit python3.4. The python2 series is troublesome to handle Japanese, but the python3 series can handle Japanese simply. Also, in order to analyze data, modules such as numpy are required at the top, and if you try to install them individually, you will suffer from dependencies, but with anaconda you can install them all at once and uninstall them easily. You can also pip in anaconda environment. I was able to move mecab as well. anaconda is also great. Download Anaconda
Machine learning This is scikit-learn. There are many ways to do machine learning, but it doesn't work, so I focused on scikit-learn, which seems to be of good quality.
Editor and execution environment The editor is gedit. For windows, it's Sakura Editor. Write the code in an editor, paste it into the qtconsole and execute it. qtconsole is also great. Since it is ipython, tab completion is also possible. This completion and help function is really useful. You can explore the vast amount of python functionality as needed.
I also used IPython notebook, which I think is good for communicating information, but it was difficult to use. Especially, I found it difficult to use to display the graph inline. The code is hard to see. qtconsole is easier to use.
I have spyder in my IDE. It's well done, but it was hard to use. It may seem easy to use when developing software, but I didn't need it.
Recommended Posts