62

                                            That's actually a pie chart out of AccountBlaster,

                                    but this is from Seaborn, from our class, plotting mass data with NumPy, Pandas, Seaborn, & MatPlotLib & a little color addition in The Gimp.

These are some crazy powerful tools we are learning, on several fronts, going back over and reviewing statistics and probability which is essential for analysis, betting, all kinds of things, and the great mass-number crunching tools NumPy, Pandas, and the graphing plugins Seaborn, MatPlotLib and how to do all kinds of super powerful mass-number-crunching tricks with these tools that I would never know how to do without this class.

I love the measles problem, first we generate 10,000 people, then 5% of them get measles, 5% of those test negative even with measles, and 5% of the no measles people falsely test positive for measles even though they don't have it. We simulate actual randomized numbers with NumPy and Pandas and Python and then do the equivalent with just Bayes Law and algebraically manipulating the probabilities to make sure we understand Bayes Law, which is super interesting now that I understand it. I have been doing these kinds of experiments since I was a kid, but NumPy is optimized for experiments like this and so these programs are tiny but do an immense amount of computation with just a little tiny bit of programming.

For example say you have a two-dimensional DataFrame df, consisting of a grid of 9,875,492 rows and 197 columns of various numbers. What's the average of every column? Well, you can make a double loop, or, you can use NumPy & Pandas optimized vectorized operational capabilities and just say df.mean() !!

The tools are really amazing and I am extremely excited to be able to learn to start acquiring them. They are vast but also simple so it's both overwhelming while you can build various graphs pretty quickly and in a surprisingly short time for the jumble of data in the program..

That graph is out of Seaborn, from our class, but actually I did doctor the colors up in The Gimp. I think you can do that with the software too, but I'm far from proficient so I'm still working on learning how to change all the colors. I'm seeing hue='Category' in some of the lines of code we are running, which is what gave me the idea to change the color hue, because it didn't work how I thought it would in the software - but I bet you can make the software do that too, I'm just not exactly sure how yet.

Why are pie charts frowned upon? No other kind of graph can show that clear of a visual representation of a set of data where everything adds up to 100%, like a pie chart can.