I am currently working through Python for Data Analysis by Wes McKinney. It’s a really good introduction to doing data analysis in Python using a library he developed called Pandas.
The book uses IPython so that you can work through the book in your own IPython notebook. You can start IPython with Pylab which imports Numpy, Scipy and Matplotlib. You can also set the display to inline so that you can see your plots drawn.
Why would I want to build a skillset in data analysis with Python? Because R and MATLAB are simply not good enough for production deployment in a quantitative finance setting.
Python is ideal for research and for non high frequency trading production deployments of trading algos and the like, ideal for deployment too.
If you need to improve the performance of some modules you can push that down to something in C or C++ that is closer to the machine.
But for my purposes at the moment – learning about quantitative analysis and building my own macroeconomic models – Python is pretty close to perfect.
If you desperately need to use R code because that’s what you have a lot of experience with you can use R libraries in Python and not repeat yourself!
Once I’ve worked through the book and played with large domestic datasets (RBNZ spreadsheets and large time series from Stats NZ) I plan to share some of my research with you. I’ll have to work with statsmodels but through embracing iterative development I’ll start with basic macroeconomic models.
My medium term goal for this research project is to build an NZ economy model that incorporates sentiment analysis distinct from current business confidence surveys.
Why distinct from current business confidence surveys? Because I have a suspicion that the current business confidence surveys do not cover a broad enough sample of business owners to reasonably gauge business confidence. Having looked at the “companies and organisations spoken to” in different surveys, I believe there is room to leverage data driven sentiment analysis to build a better “confidence” indicator.
That’s a few projects away but should give you an idea of where I’m aiming with this line of research.