These are chat archives for FreeCodeCamp/DataScience
discussion on how we can use statistical methods to measure and improve the efficacy of http://freeCodeCamp.com
@Evaderei typical exercise! Nice introduction, I absolutely invite you to try. Random Forest is very robust, that's the reason
The only thing is: the 90% of us who participated in kaggle have tried Random Forest without really understanding what other methods can do, or the need of feature creation/transformation. It is essential to investigate more about that and other methods and aspects of the analysis cycle.
The Survey 2016 dataset is available!!!
This week we did more work on the Survey 2016 and started working on the final dataset (@erictleung and co.), with reports by @SamAI-Software and me (@evaristoc). Some draft for visualizations in d3.js were also discussed. Because the purpose is using d3.js for the project, we were also collecting references about d3.js: best practices, usual programming patterns, and examples, just to get inspired.
The analyses of the dataset have been carried out using R, but other tools are welcome.
For those with an interest in Machine Learning, @zcassini and other people have been providing links about free-access books and courses about the topic, and discussed things like Support Vector Machines, kaggle, Multivariate Statistics with some mention of Linear as well as Multilinear Algebra.
If you want to contribute with visualizations about the Survey 2016 dataset, please keep checking the repo and this list:
List of Week Contributors: profoundhub, krisgesling, bitgrower, evaristoc, gitdek, QuincyLarson, erictleung, ConfusedRunningGuy, gayathry2612, jinhuaxiao, Evaderei, ddrdushy, emilaasa, SamAI-Software, zcassini, s-marcelle, koustuvsinha, jacobbogers, mmacedoeu.
Announcement: QuincyLarson Survey 2016 finally available
Some Links and Popular messages:
DSR (DataScience Room) is an effort to gather campers of all levels and specializations willing to engage in discussion, collaboration and practice of data-related projects, with preference for using FCC data.