How well do Python machine learning algorithms scale?
Question:
How well do Python machine learning algorithms scale for web applications?
So far, I am impressed with the Mahout library but as a Python programmer I was hoping there are also libraries catered towards scaling really well.
Answers:
Pure Python code tends to run slower than compiled languages. Instead, it favors rapid development.
When Python coders need speed-ups, they tend to use C/C++ libraries with Python wrappers, use NumPy, run pypy, or construct fast extensions using Cython.
These alternatives could each be applied to a machine learning application. A couple tools that are popular (and seem to scale well) are PyBrain and PyML.
How well do Python machine learning algorithms scale for web applications?
So far, I am impressed with the Mahout library but as a Python programmer I was hoping there are also libraries catered towards scaling really well.
Pure Python code tends to run slower than compiled languages. Instead, it favors rapid development.
When Python coders need speed-ups, they tend to use C/C++ libraries with Python wrappers, use NumPy, run pypy, or construct fast extensions using Cython.
These alternatives could each be applied to a machine learning application. A couple tools that are popular (and seem to scale well) are PyBrain and PyML.