Friam

Stanford Modeling Class

Classic

List

Threaded

3 messages Options

Owen Densmore

Stanford Modeling Class

Administrator

The Stanford Modeling Class has started, I thought I'd give a summary of what's up so far. The website is: https://www.coursera.org/modelthinking/

First of all, this is NOT a deep dive into exotic techniques. Rather it is a *very* broad overview of modeling, answering the question "why model". With each discussion point Scott gives concrete examples, but without having to write code or "do the math".

The name of the class, Model Thinking, captures this difference: he is guiding us through a new way of thinking that is precise and relatively well understood by now.

So it is much more a very high level view of modeling (mainly Agent Based Modeling but also simple mathematical and graphical models) with the emphasis on very clear thinking.

One quick example: Aggregation. This is the reductionist dilemma. How do you either

1 - Look at a Macro event and deduce its parts, or
2 - Look at simple Micro rules and deduce the results.

Water, a micro molecule and a macro substance. The molecule cannot be "wet". Or Schelling's segregation model: at the micro level, individuals are quite tolerant, wanting only a few like neighbors, yet the result is a surprising large value of segregation. He also introduces a metric for segregation, The Dissimilarity Index, so we can be precise. He also looks at the Game of Life and CA's in a similar way.

Unlike the Machine Learning class, there are readings, generally classics in the field. The first session's readings, for example, are Josh Epstein's "Why Model" and Scott's introduction to his class. Both are very "humanities" over "computation".

I've uploaded my class notes, 2 2/3 sets thus far: they are screen captures with pdf annotations. You can get a feel for the class quickly by thumbing through them. They are at http://backspaces.net/temp/ and begin with ModelThinking.

I do have to confess: there is a method in my madness in writing this email. I find learning in a cave, by myself, less fun than having some others along for the ride. So if anyone does take the bait .. and ends up following the class, lets get together and chat about it. And don't feel you have to be a Scientist or Mathematician or Hacker .. you don't.

-- Owen

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

Russ Abbott

Re: Stanford Modeling Class

HI Owen,

My wife and I (she teaches Renaissance English and is definitely a Humanities person) are "taking" the course together. My Google+ comments are here and here. I'm not certain what the structure of the course is. The lectures all seem to be uploaded, but I haven't seen anything about a timeline, exercises, tests, or any other structure for the course. Do you know anything about that?

-- Russ Abbott
_____________________________________________

Professor, Computer Science
California State University, Los Angeles

Google voice: 747-999-5105

Google+: https://plus.google.com/114865618166480775623/

vita: http://sites.google.com/site/russabbott/
_____________________________________________

On Tue, Feb 28, 2012 at 9:14 AM, Owen Densmore <[hidden email]> wrote:

The Stanford Modeling Class has started, I thought I'd give a summary of what's up so far. The website is: https://www.coursera.org/modelthinking/

First of all, this is NOT a deep dive into exotic techniques. Rather it is a *very* broad overview of modeling, answering the question "why model". With each discussion point Scott gives concrete examples, but without having to write code or "do the math".

The name of the class, Model Thinking, captures this difference: he is guiding us through a new way of thinking that is precise and relatively well understood by now.

So it is much more a very high level view of modeling (mainly Agent Based Modeling but also simple mathematical and graphical models) with the emphasis on very clear thinking.

One quick example: Aggregation. This is the reductionist dilemma. How do you either

1 - Look at a Macro event and deduce its parts, or
2 - Look at simple Micro rules and deduce the results.
Water, a micro molecule and a macro substance. The molecule cannot be "wet". Or Schelling's segregation model: at the micro level, individuals are quite tolerant, wanting only a few like neighbors, yet the result is a surprising large value of segregation. He also introduces a metric for segregation, The Dissimilarity Index, so we can be precise. He also looks at the Game of Life and CA's in a similar way.

Unlike the Machine Learning class, there are readings, generally classics in the field. The first session's readings, for example, are Josh Epstein's "Why Model" and Scott's introduction to his class. Both are very "humanities" over "computation".

I've uploaded my class notes, 2 2/3 sets thus far: they are screen captures with pdf annotations. You can get a feel for the class quickly by thumbing through them. They are at http://backspaces.net/temp/ and begin with ModelThinking.

I do have to confess: there is a method in my madness in writing this email. I find learning in a cave, by myself, less fun than having some others along for the ride. So if anyone does take the bait .. and ends up following the class, lets get together and chat about it. And don't feel you have to be a Scientist or Mathematician or Hacker .. you don't.

-- Owen

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

Owen Densmore

Re: Stanford Modeling Class

Administrator

On Tue, Feb 28, 2012 at 7:58 PM, Russ Abbott <[hidden email]> wrote:

HI Owen,

My wife and I (she teaches Renaissance English and is definitely a Humanities person) are "taking" the course together. My Google+ comments are here and here.

Nice use of G+ .. and I agree that the first module was a bit light. But the second and third were fine. Tomorrow I'll start in on the forth, Decision Theory.

I'm not certain what the structure of the course is. The lectures all seem to be uploaded, but I haven't seen anything about a timeline, exercises, tests, or any other structure for the course. Do you know anything about that?

I think all the coursera classes are having a slow start compared to last fall's sessions. Scott has done a good job of getting the videos available, and the readings as well. The quizes are likely a tech problem. It is odd there is less structure in terms of what is expected from the student.

In the ML class, the quizes (called Reviews) were really well done, randomized so you could take them often as you'd like. There was a quiz (5 multiple choice questions, often with 4 parts to the question) for each section, so two a week.

And in terms of structure, you were given time lines and due dates in the ML class. I imagine coursera is the culprit there too.

I suspect Scott is trying for that and its running into problems. I don't think there will be programming exercises, while the ML class had extraordinary programming exercises, artfully integrated into a teaching script. Each programming exercise had around 6 programs the student wrote, then submitted for automatic testing and grading.

-- Owen