A good visualization can help a lot of feature engineering. Even with deep learning if you can through in human insight in the mix it might make it easier. I understand that it is lot of practice but still a course that goes a practical way will be really helpful. Does anyone has any suggestions for a data visualization course?
I think a lot can be done with matplotlib.pyplot and pandas. Probably more important to get very familiar with one data visualization library, would help you get more creative and iterate faster. There are loads of data visualization courses on the popular online course websites, easy to get overwhelmed.
Any of those that anyone may gone through that feels practical and gets you a good start? I guess it’s not about the tool but about the techniques.
The various libraries tend to have good examples.
Check out the galleries for Matplotlib, Seaborn, and Bokeh and try to understand the code examples.
You can also check out Mike Bostock’s work (creator of D3). A lot of the cooler stuff out there is inspired by/taken directly from him. https://bl.ocks.org/mbostock
I did the one by John Hart (UIUC) a while back:
I thought it was good, because it gave some general principles about visualizations. It’s not a good place if you want to learn about some particular software, though.
@msp Is it practical? I don’t really care much about the tool if it is practical. Which tools do they actually use for this?
It’s rather theoretical. The course teaches about experimental results on which types of graphics best convey certain types of information, which color schemes work best, etc. As far as I can remember there is no specific tool being used preferrentially.