How do I come up with a topic for my Honors Undergrad Thesis in Stats?

Flip a coin

>>8490071
Just drop out if you can't do it and become a codemonkey if you learned some CS

File: 1479794290205.jpg (75KB, 800x800px) Image search: [Google]

75KB, 800x800px

>>8490071
>"A Differential Analysis of The Quantitative Function of Obfuscated Semitic Influence Upon The Mainstream Media".
Okay, OP?

>>8490071
Where can I find a girl like that, /sci/?

>>8490851
Pretty much any union of feminists, anon.

>>8490855
can't say I'd be against dating a "feminist" if they looked like that

>>8490861
What if it was non-binary and liked to remind you that you must have written consent before grabbing her by the pussy? It also must be signed by her and yourself, dated and then collated and stamped by her transexual non-binary male identifying toasterkin attorney (who is her 'ex'-lover).

>>8490861
>>8490865
Correction: zhir by the pussy?

>>8490865
no one could be that fucked up without having some color of dyed neon short hair

>>8490879
I guess.

>>8490865

Tumlberinas really should start using god as a pronoun. It'd be a lot funnier.

>Social Justice on Social Media: Text Mining to Forecast the Incidence of Radical Progressivism, with a Machine Learning Model to Identify Individuals Susceptible to Leftist Radicalization

But seriously, stats is one of the most far-reaching STEM disciplines, so you can start by picking a research question that interests you and for which there is data available, and make that your one-year project.

File: succeed-others-must-fail.jpg (58KB, 497x470px)

58KB, 497x470px

>>8490071
If you want to do something useful. analyze some academic papers and find errors in them. Bonus points if you can get them retracted or at least corrected.

Protip: social 'sciences' are the worst, but medicine is almost as bad, and more important.

File: 99iLHrx.jpg (199KB, 625x461px) Image search: [Google]

199KB, 625x461px

>>8490071
do it on information theory, it will change the way you think of statistics.

heres a fun project: take a series of predicted temperatures and a series of measured temperatures, then estimate mutual information between the prediction and the measurement.

you can test the hypothesis:
> 'the further out the predictions are, the less bits of genuine information they give us about what the actual temperature will be'

there is a lot of background info you can write about
> wtf is entropy clod shannon etc etc
> wtf is mutual information
> how do you interpret these quantities
> how do you estimate them in practice
> how do you demonstrate that these estimates converge
> how much data do we need to expect a reliable estimate


if you can't tell i'm biased towards info theory

You should have an advisor. You'll do your thesis on something your advisor studies and can advise you on.

>>8490071
>A revolutionary simplification of statistical mathematics: Either it happens, or it doesn't

File: Trump_Victorious.jpg (360KB, 633x568px) Image search: [Google]

360KB, 633x568px

>>8490071
Easy.. study why polls failed to predict Trump victory even up to the evening before the election.

>>8490071
who dat gurl anon?

>>8491645
Isn't pic related a Gore Vidal quote?

>>8492036
>why polls failed to predict Trump victory
because people lied about what they were going to vote