Anyone gone through the process of learning tableau or R?

ByAnyMeans

Rookie
Joined
Nov 18, 2016
Messages
197
Reputation
110
Daps
457
I would say R hands down. What ever you want to do in tableau can be done in R and shared and reproduced easily. The r community is very robust and I think you won't have to get super deep in the stats to spit out some charts and stuff. But you will have to understand core programming principles to do meaningful stuff. Core being looping, and conditional statements. Not hard stuff but that combined with the nuance of r and it's syntax for someone nontechnical could be hard. But once over the learning curve your good.

Only thing is you need a lot of data in the workplace and time for it to prove valuable. I think everybody should know programming so go for it.
 

Brown_Pride

All Star
Joined
Jun 8, 2012
Messages
6,416
Reputation
785
Daps
7,886
Reppin
Atheist for Jesus
If you understand excel at an intermediate level, meaning you can nest if then statements, do vlookups, maybe even know what an array is then tableau is cake. Either way though you should be able to learn it with some fiddling around in it. Not familiar with R.
 

Prevail

Pro
Joined
Nov 20, 2016
Messages
200
Reputation
-40
Daps
553
Reppin
Somewhere
And this is coming from someone that uses Python+pandas 90% of the time.
I know numpy / scipy / pandas / pyqtgraph etc stats tools might not be mature enough for you.
Do you really believe python won't supersede R in stats departments in the next 10-20 years once these libraries mature.
The Machine Learning libraries are already becoming well known.
Tensorflow scikit-learn etc are pretty well known.
 

badhat

Pro
Joined
Jun 3, 2014
Messages
567
Reputation
228
Daps
1,775
I know numpy / scipy / pandas / pyqtgraph etc stats tools might not be mature enough for you.
Do you really believe python won't supersede R in stats departments in the next 10-20 years once these libraries mature.
The Machine Learning libraries are already becoming well known.
Tensorflow scikit-learn etc are pretty well known.

Tensorflow? How is that going to have an effect on Python superseding R in stats depts? Deep neural nets are fun to play with, and they have a place with irreducibly complex questions, but they don't lend themselves to statistical inferences — you get a completely opaque process. The more accurate they become with additional layers, the harder it is to make the case that x->y.

As for scikit-learn, I think the new hotness is xgboost and maybe lightgbm, a set of boosted decision tree libraries. I don't see anyone really pulling for support vector machines anymore. sklearn has linear models, but nothing as good or as useful as R's glm. And if I have to grab an ANOVA, it's literally more straightforward in Excel than it is if my information's in a pandas dataframe.

This whole "machine learning" craze is getting out of hand. If you're modeling for business or logistics, you want actionable reports. A potential boss can be informed by my linear model, even if the r value is .7. A report based off information I made with a sexy tensorflow process is going to end up with me standing at his desk, mumbling that I don't actually know why A and B cause C, but the computer ran the simulation 100K times, so…
 

badhat

Pro
Joined
Jun 3, 2014
Messages
567
Reputation
228
Daps
1,775
90% of the stuff you can do with R, you can do with Python. If you already know Python, I'm not saying you need to drop everything and grab R.

If I have a dataset that I'm suspecting follows a Poisson distribution, and I want to check that, I'm importing one library in R, ggplot2, and that's not necessary, it's just cause I like how it looks. I can do everything I need without importing another library, nls is right there.

For Python? I'm bringing in pandas to get my data into a dataframe, I'm importing matplotlib.pyplot for plotting. That's bringing in numpy on its own. If there's categorical factors, I'm just going to bite the bullet and use seaborn, cause holy shyt, pyplot makes multicolor more tedious than it has to be. Then to actually make the fit, I'm importing statsmodels' poisson fit.

Can I do that? Sure. It's just more of a pain in my ass when pandas updates and decides to depreciate a method I was using, or I can't find a function in statsmodels because it's actually in scipy.stats, and I'm just misremembering.
 

TLR Is Mental Poison

The Coli Is Not For You
Supporter
Joined
May 3, 2012
Messages
46,179
Reputation
7,465
Daps
105,756
Reppin
The Opposite Of Elliott Wilson's Mohawk
I taught myself both this year. I know just enough in R to do some linear regressions and p-tests. I mainly use Tableau for geoplotting. Thing with both is, at least in my experience, you really need to know what the hell you want out of them to make any use of them. Might just be me but I learn how to use a tool best when I have a goal in mind.
 

King Kreole

natural blondie like goku
Joined
Mar 8, 2014
Messages
13,864
Reputation
4,308
Daps
39,699
I taught myself both this year. I know just enough in R to do some linear regressions and p-tests. I mainly use Tableau for geoplotting. Thing with both is, at least in my experience, you really need to know what the hell you want out of them to make any use of them. Might just be me but I learn how to use a tool best when I have a goal in mind.
How much programming background did you have before teaching yourself R?
 

Serious

Veteran
Supporter
Joined
Apr 30, 2012
Messages
78,759
Reputation
13,986
Daps
186,406
Reppin
1st Round Playoff Exits
I tried fukking with R a couple months back, but didnt have the time to put into it. I have no real point of reference to start from from, but I have strong background in using programs like SPSS and SAS....
 

The Minister

Look at your face
Supporter
Joined
Feb 20, 2014
Messages
51,484
Reputation
12,801
Daps
195,944
Reppin
Handsome Boyz Ent
Tableau is pretty intutive/self explanitory assuming you have basic computer skills. That said i havent messed with that stats ish since undergrad
 

EndDomination

Veteran
Supporter
Joined
Jun 22, 2014
Messages
30,915
Reputation
6,976
Daps
108,493
Used R for statistical analysis, as a sort of secondary to SPSS.
Showed my brother how to use it for Oncology-Research, there are beginners books w/ online data sets that you can teach you the basics pretty easily.

Never tried Tableau.
 
Top