These are chat archives for FreeCodeCamp/DataScience

Sep 2016
Sep 12 2016 20:08
anyone into Topological Data Analysis???
Sep 12 2016 21:53

@albert2309 Hi man! yes, the point made in the book is very interesting - it is more about the wide spread misuse of analytics without a clear understanding of why to use different kind of tools...

@erictleung @albert2309: the problem suggested by the book is unfortunately not new and it is more concerning nowadays. Have you ever hear the quote: "There are three kinds of lies: lies, damned lies, and statistics."? It was popularised in USA by Mark Twain back to 1900's but it is believed to be older than that...
You can see some back use of those tools in every election period ;)

@ascheist Interesting... reading about it just now: I heard of it just once, in a meetup with one of the founders of Apache Spark (they seem to be working on something there in Berekeley...), but never since... Are you doing some? Can you comment about your experience? Topological approach seems to be more mathematical, with strong use of algebra and with a lot of applications in time series analysis. My background includes classic frequentist statistics mostly, although I have been studying modern approaches lately, particularly bayesian and then ML, but never used topological approaches before. Any good reference? Python or R?

Sep 12 2016 22:02
@ascheist reading about its application in statistics... for some cases makes a lot of sense... also in the case of classification... I think topological analysis could be a way to re-evaluate the way we normally distinguish local vs global optima in a "persistent" topological space... Interesting...