Statistical Analysis of Beer Drinking

Still inputting all my beers reviews but I've been working on this:

https://public.tableau.com/views/BeerReviews/MyBeerStats?:embed=y&:display_count=yes&:showTabs=y

 

How far off the rDev are your ratings as a whole? Are you over ratings a lot of stuff? Seeing 3.87 (I tend to mark 4 as my "will buy again" #) as your average rating can mean a number of things.

Is your most rated style also the highest rated? The backbone of BA is all #s really. I love reading and seeing this stuff. Thanks

 

I think breweries and or distributers would like info like this

 

Yay! Numbers! This is fun! Thanks for sharing!

Some other things I'm curious about:

• How many of these 2,000 unique beers were barrel aged (if any)? What was your average rating for this subset of BA beers (compared to your overall average rating)?
  
• How many of these beers were the result of trades? In which states were the traders with whom you traded most often?

 

Lots of angles to go as you analyze

Just quickly glancing at your profile here. Of the top 50 rDev scores that you gave, quite a handful are pils,euro lagers, or bocks. It likely shows what you already know. Maybe BA raters are underrating the pils/lager category. Maybe you like that style more than the "norm". Are you finding that you are trying less of what you really dislike, and more to keeping with your "numbers". I try to be an open book every time, but its not easy. I don't want to be the type that says "I've tried 20 types of berliners. I won't even give this a shot". It can lead to us playing favorites or what we've become familiar with ie (rating, sampling, and possibly over rating IPAs as a whole).

 

This is cool. Are you inputting your stats manually or is there some way to export from BA that I haven't played around with?

 

My analysis is pretty much limited to calculating the chance I have at the girl I'm drinking with.

 

An uber geek thread in an already Geeky forum

A round of Bayesian Bavarian Standard Deviation Dunkel Doppelbocks for all !!!

 

That's awesome, I wish I could do that for the beers I have had, it would be interesting......

 

The calculation I do when drinking high ABV beers, particularly when I need to drive home, is to determine the comparable number of "normal" 4.2 ABV beers. Multiplying 12oz times 4.2 ABV gives you 50.4, representing a normal beer. When I order a high ABV beer, I multiply the ounces by the ABV, and then divide by 50.4, giving the comparable number of normal beers. So 20oz of a 10.1 ABV beer is equivalent to about four normal beers. Probably overthinking this, but if you're driving it gives a person a better gauge of how much they've had since we all figured out in high school how many normal beers we can handle.

 

Damn it @#%!!!
I just got home from work, and now you are making me do more work. I'll get back to you with the results.

 

I'm most impressed with the total of 550 beers sampled from Minnesota. time for a book, "The Beers of Minnesota".

 

There are more interesting things you could find out.
With a simple Excel formula you could calculate the standard deviation, and then use that to identify outliers, which are usually defined as being 1.5*SD away from the mean.
A slightly more complicated but still interesting problem would be calculating a prediction interval for a new beer rating, which would let you identify beers whose ratings are exceptionally rare.
I will have to start keep track of these things for myself!

 

So - at one point in time, I was able to export my reviews via a solution developed by another BA, but this functionality doesn't work anymore - most likely due to the number of hardware and security enhancements that haven been happening over the years. I have to manually enter the remainder of my reviews which is why this is a work in progress. I've suggested on a few occasions offering the ability to purchase exports of your data from BA - I'd pay for the convenience!

 

Congratulations! As far as my personal stats go, I would love more style related info like under IPA it should say how many I have had, and then rank them according to my rating. It should also say if that beer is a top 250 or listed as an exemplar of it's style in the bjcp. Wishful thinking I know. =)

 

I LOVE the software (is that the correct term?) you used to create this! Does this have a tutorial to help me if I wanted to create the same? P.S.: I'm a relatively "older dude" whose computer knowledge is less than stellar

 

Beer nerd stuff....love it! You are much more dedicated to the statistical data keeping of your drinking than I am for sure. I finally started using untappd because I have a horrible memory, so when I go to my favorite watering hole with 100+ taps I can remember what I have had or not (and liked or not). Well done!

 

So I'm logging my ratings in a google spreadsheet. I'm a statistician, and I'll say that analysis of beer ratings going beyond very basic stuff is tricky, because each beer experience influences future experiences (particularly within a style or group of similar styles) and bias in ratings based on style (as mentioned by others above). What I intend when I get enough ratings to do is to try to identify my favorite brewery using a mixed effects model, where a grouping of similar styles is a random effect.