Valuation Engineer Hub - Plans

[RCA]

I already have a https://valuationengineer.com OJS journal, - still empty for now. It is stored on Hostinger in Lithuania: Fast and relatively inexpensive.

To this website will add "Substack" as a gateway to technical articles on valuation. Substack will link into a Quarto subsite which will allow R, Python code to run. In fact I will be able to link to Jupyter, which supports 40 languages. So I am sure C++ and perhaps Prolog will also show up. There will be links to GitHub to download code, links to YouTube or other videos with tutorials.

There will be plenty of math. MARS (multivariate adaptive regression splines), Neural Networks and other types of programming will be discussed with sample code and data.

The site will be strictly technical. There could be a forum similar to AppraisersForum - but only for advanced technical issues.

There probably eventually be a fee of $5.00/month via Substack because that is the minumum and I want to know who I am dealing with.

Users will be able to request that I cover or explain certain technical issues.

There may be non-appraiers who are also engaged in MARS (R/earth) work, but it will in any case be restrict to technically inclined members.

 

[Tom D]

I'm not going on the spaceship.

 

[RCA]

Take my word for it, trying to install Jupyter on a remote Linux server is a *****. .... But, I got it to work. However, there are other considerations:

1. GitHub "Launch Binder" is better for users to execute the code I give them, and is free.
2. Multi-user Jupyter is too expensive and has plenty of security issues. Especially in Europe, which is very strict regarding privacy (in shared Jupyter, which is cheap, users can clobber each other's code as well as see others' code).
3. So, I will create Quarto R or Jupyter code for GitHub with a button to allow users to execute for free in 2-hour time slots.

GitHub is reportedly safe and provides privacy. The only downside to GitHub's Binder is that after 2 hours, your code disappears and you have to reload from scratch. The best thing to do is install RStudio on your own computer and download the code (more complicated, of course).

 

[RCA]

Note: I have Jupyter and Quatro and RStudio running from my Journal via Substack. Click on the Substack menu item and that will take you to:



'Executing an R Earth Program in Jupyter' - actually brings up an instance of RStudio with the R code. And, you can actually execute it, modify it and execute it again. First time you run it is is a bit slow. But this is mostly for beginners. If you want to do a lot more R - then it is better to download your own R and RStudio.

Because I can get Jupyter to run - I can also teach how to use many other languages. My next steps:

1. Make sure I can execute Python for some really simple neural network processing (like identifying a cat in a photograph.
2. Add Prolog.
3. Two new appraiser articles in the making - but one will not be of much interest to this audience: Algebraic Structures Applied To Appraisal. ....
4. Protcol Engineering
5. More on the RCA method
6. More realistic housing data, with the ability to let you run R earth, modifying the hyperparameters to learn how they affect the quality of the models.
....

I have toooo much work to do.

I plan on creating demos. Subscriptions for the few that want help on using MARS for their appraisals.

 

[Tom D]

UFRO - unidentified flying regression object. Earth to mars. Mars attacks was a great movie.