Machine Learning, sources of information

My problem

I want to learn Machine Learning concepts, understand how to apply on real cases.

There are so many sources of information, some of them with good/bad quality, some others very complex.

The questions are simple:

  • when is useful to use machine learning solution?
  • what is a neuronal net?
  • what are the steps to build a machine learning solution?
  • what are the type of algorithms you can use for specific problems?

The solution I found

  1. I’m doing some Skillsoft courses I have available in the company platform. These are covering the basis and now I’m jumping on specific topics. They are nice as they include examples in Python with scikit.
  2. Review these notes from @TessFerrandez: Notes from Coursera Deep Learning courses: https://www.slideshare.net/TessFerrandez/notes-from-coursera-deep-learning-courses-by-andrew-ng
  3. Watch the video from @TessFerrandez: Machine Learning for Developers . The video explains how the process of building a machine learning solution is, explained in plain English and with very nice examples easy to remind. The video helped me to link a lot of technical ideas explained in the courses with a natural flow.

 

Pydev with Eclipse

I find EMACS too much complex for me. I want to concentrate on the analysis of data with Pyhton, so I have looked for and alternative IDE for Python: Pydev.

The steps I have followed are:

  1. Download and install Eclipse (in reality I had it on my computer).
  2. Then, download, install and test Phyton.
  3. Once done, download and install Jython.
  4. Download and install IronPython.
  5. Install PyDev.
  6. Create a new PyDev project.
  7. I have created a hello.py file with print(“hello world”), then I have debugged it and it worked.
  8. Create a package to gather all the information.
  9. The set-up of the environment is very well explained. You first have to configure the interpreter, then you can develop all you need. In my case I’m starting importing some finance data from Yahoo Finance.
  10. Using Windows>Properties>Python Interpreter , install library “numpy” using “pip” as installer. This step is crucial to me as Python interpreter just work on PyDev in my computer.

EMACS

I’m sure EMACS runs faster and it’s more efficient than Eclipse, but the point is that now I can concentrate on the development of the scripts I want to implement. 🙂

Zipline, for trading simuation

Zipline is an open-source algorithmic trading simulator written in Python.

I have installed Zipline library with pip, in my case, as I’m using PyDev I went to Window>Preferences>Python Interpreter>Install/Uninstall with pip

Once I have tried to use it, I was not able to find the quick way to build the basic code I wanted, so I finally has uninstalled it as with numpy and pandas I have enough.

So I was not able to understand the power of this library, by the moment I do not need it.