Towards thearetical understanding of deep learning by Sanjeev Arora [outside article]

|

A presentation on Deep Learning including a brief history and tutorial.

https://www.dropbox.com/s/qonozmne0x4x2r3/deepsurveyICML18final.pptx

Which test to use in what situation [outside article]

|

Having trouble deciding what statistical test to use for your data? Use this handy flowchart from Penn State to decide. It includes a review of all the statistical techniques provided, as well as a table consisting of inferences, parameters, statistics, types of data, examples, analysis, Minitab commands, and conditions.

 

 

https://newonlinecourses.science.psu.edu/stat500/node/67/

American Statistical Association’s statement on p-values [outside article]

|

The American Statistical Association (ASA) has released a “Statement on Statistical Significance
and P-Values” with six principles underlying the proper use and interpretation of the p-value. The ASA
releases this guidance on p-values to improve the conduct and interpretation of quantitative
science and inform the growing emphasis on reproducibility of science research. The statement
also notes that the increased quantification of scientific research and a proliferation of large,
complex data sets has expanded the scope for statistics and the importance of appropriately
chosen techniques, properly conducted analyses, and correct interpretation.

Read more:

https://www.amstat.org//asa/files/pdfs/P-ValueStatement.pdf

Why Frank Harrell does not like p-values by Frank Harrel [outside article]

|

With the many problems that p-values have, and the temptation to “bless” research when the p-value falls below an arbitrary threshold such as 0.05 or 0.005, researchers using p-values should at least be fully aware of what they are getting. They need to know exactly what a p-value means and what are the assumptions required for it to have that meaning. ♦ A p-value is the probability of getting, in another study, a test statistic that is moreextreme than the one obtained in your study if a series of assumptions hold. It is strictly a probability about data, not a probability about a hypothesis or about the effect of a variable.

Read more:

http://www.fharrell.com/post/pval-litany/

Note on small p-value hacking by Thomas Lumley [outside article]

|

The proposal to change p-value thresholds from 0.05 to 0.005 won’t die. I think it’s targeting the wrong question:  many studies are too weak in various ways to provide the sort of reliable evidence they want to claim, and the choices available in analysis and publication process eat up too much of that limited information.  If you use p-values to decide what to publish, that’s your problem, and that’s what you need to fix.

http://notstatschat.tumblr.com/post/172190899051/small-p-hacking