George Dell - Stats, Graphs And Data Science

[Joyce Potts]

I have heard nothing but great reviews about this class. Any additional comments??

https://georgedell.com/sgds1-orlando-fl/

 

[Renee Healion]

If you go let us know. I like his email newsletter but have taken no class since he seems to hold them far from here.

 

[glenn walker]

He is a nice guy that's all I will say - Take it and let us know what you learned !

 

[hastalavista]

I know George; unfortunately, every time he has his class up in my area, I have a conflict in my schedule, so I have not taken his classes (yet).
I know a lot of people who have taken his class (repeatedly); I cannot recall any saying anything negative about the class. I know quite a few who swear by what they learned.

 

[leasedfee]

Take his class. I haven't taken the class; but I want to. Super smart gentleman and passionately knowledgeable about appraisal and statistics.

 

[Elliott]

I took his class on adjustments a half dozen years ago. Better than a USPAP class. Best insight was into his view adjustments criteria. He's probably more of a statistics wonk than appraiser's should be, but the class was good for how to 'talk about adjustments' even if the market isn't up to the precision of the data.

 

[Rlong]

He's also teaching this course next week in Nashville, and June 7 & 8 in Indianapolis. I am thinking of attending in June. It looks like he is using R open source software. Has anyone seen or used this for regression analysis? It might be more universal to learn in Excel, I’m assuming there are reasons he works in R.

 

[leasedfee]

universal to learn in Excel, I’m assuming there are reasons he works in R

R is a dedicated statistical program. Excel, even using the built-in free Multi-Variate Data Analysis package, is very primitive. In Excel, the stat outputs are limited, like no Confidence Intervals, Prediction Interval, or Leverage calcs. It's time consuming and error prone to calc those by building a formula. You have to layout the columns just-so. If you want to transform a series to non-linear variables, like ln(x) or 1/x, you have to do it manually. It can't do step-wise regression where the computer does many many models to find the best fitting one. It can't handle missing data. It can't do cluster analysis, survival analysis, time-series models, or Logit/Probit models, where you scale probabilities between Yes/No, 0/1. You can buy some Excel visual basic add-ons but at that point you might as well use a real program.

 

[Artemis Fowl]

R is a dedicated statistical program. Excel, even using the built-in free Multi-Variate Data Analysis package, is very primitive. In Excel, the stat outputs are limited, like no Confidence Intervals, Prediction Interval, or Leverage calcs. It's time consuming and error prone to calc those by building a formula. You have to layout the columns just-so. If you want to transform a series to non-linear variables, like ln(x) or 1/x, you have to do it manually. It can't do step-wise regression where the computer does many many models to find the best fitting one. It can't handle missing data. It can't do cluster analysis, survival analysis, time-series models, or Logit/Probit models, where you scale probabilities between Yes/No, 0/1. You can buy some Excel visual basic add-ons but at that point you might as well use a real program.

R is open source? As in Linux based free?

https://www.r-project.org/

Answereded my own question.

 

[Rlong]

Thanks Tim! (leasedfee) that was what I was kind of surmising, I did sign up, the CORRECT dates for the Indianapolis class are June 13-14. They sent me links to download "R" (reference above), "R Studio" and "Gnumeric" (open source spreadsheet). I printed the intro / manual for R so I'm at least going to try to learn the terminology/concepts prior to class. I also am reading his article from last year in the Appraisal Journal, should be a good class!