OpenOpt, got some code to "run".

I made a little progress on OpenOpt. I got some code to "run". Now I need to spend some time to find how it can give me the same results as the same model in Excel. My latest OpenOpt Post shows my current state.

OpenOpt and I started following someone's Blog

So I carved out a little more time to work with OpenOpt. I ran a built-in example named lp_1.py, since it was a Linear Programming example. By default the example script calls on a solver that is not included with the install of OpenOpt . Below is the line of code. The funny part is read the comment next to it and ponder why would this line be the default?

r = p.solve('cvxopt_lp') # CVXOPT must be installed

Any way you can see my current progress here.

On my journey to find why the example problem did not work by default, I found another blog that has similar interests to mine. Except this blogger seems to have experience in Operations/Decision Science vs. where I am just at the beginning of this journey. So I added myself as a follower to his blog: http://industrialengineertools.blogspot.com/

A good example of where we seem to be of similar interest is this article: http://industrialengineertools.blogspot.com/2009/02/does-software-hinder-innovation-in.html
I look forward to catching up with some of his other posts and future content.

Attempting OpenOpt

I decided to try using Python with OpenOpt, but I ran into a little issue with the module. You can see the bug report I filed with the OpenOpt forum. While on the forum site I found there are discussions for Linear problems, Network (mathematical not digital) problems, and Stochastic problems. These are all topics covered under Operations Management. After I get my OpenOpt install figured out, I will start with reviewing the Linear problems to see how I can build skills and participate in the forum.

Python Numpy/Scipy OpenOpt

I am still making slow progress on reading and comprehending The Science of Decision Making given various time constraints. But progress is being made. As I am building my Linear Programing (LP) knowledge I decided to seek out how I could perform LP with Python. For now, the OpenOpt library seems to be the most documented way. Here is an example of using the library for LP. Since my current book focuses on Excel, I am looking to transfer the example from the book to Python to compare the results. I know I keep repeating myself on what I plan to do, but blogging about this has the following advantages: 1) Let's you know I have not lost interest 2) Allows me to keep a document of my developing thoughts in a central place 3) Forces me to think through my plan to build operations management skills.

Linear Programmers are Sensitive

I am have been slowly making my way through the book The Science of Decision Making and also using my old Operations Management textbook to gain a deeper understanding of Linear Programming. While I have performed the graphical method of Linear Programing and have used the Solver in Excel to find an answer, I had not used the "Sensitivity Report" before.

The "Sensitivity Report" is feedback for the data model. Using it a data modeler can can compare the "shadow price" vs. the real price of an extra unit of production. If the real price is greater than the shadow price then the results of the model given is considered optimal. If the shadow price is higher than the real price of an extra unit of production then adjustments can be made to the model. Another bit of feedback the "Sensitivity Report" gives you is the range of units for each constraint before the answer would change. Hence the amount of change required to change the answer can be considered along with the shadow price to determine if the model should change and if so in which direction.

As I gain greater insight to how linear programming works, I hope to post some examples with real data. Also at this time I am using Excel as that is what both books use for examples. At some point I plan to try the same thing with SciPy's optimizer.

Ishikawa/Fishbone diagram for 2009 economic problems

A skill in operations management is to identify root cause. At this time, the US economy is struggling. With your insight, we can track the issues to the root causes for our current economic problems. The Bernoulli on Business wiki site has a Ishikawa/Fishbone diagram. Take a look and suggest what changes are required to make this more accurate of our current economic problems. I will take your suggested changes and update the Ishikawa/Fishbone diagram until it is determined complete.

The link to the diagram: http://sites.google.com/site/bernoullionbusiness/Home/projects-on-the-burner/fishbone-diagram

Traditional First Post

Hello

Welcome to Bernoulli on Business and as blog tradition leads me "First Post"!!

I look forward to the niche market of those folks who may be interested in building skills and examining issues that apply to operations management. The list of books off to the right are books I intend to review as part of building up my skills. I plan to use GNU R, Python (with the SciPy library) and Excel. If you are an open source fan you may wonder why I am using Excel. There are three reasons I am using Excel. 1) Some of the books I will be reviewing use Excel. So some of what I will be doing is building my initial skills in Excel, then expanding them with GNU R and/or Python. 2) Despite what many open source people hope, Excel is the gold standard in business for quick analysis. 3) Excel is more mature than Calc from OpenOffice.

Now you may ask, if Excel is the business standard, why are you using the other open source programs as they are not standards for business in general? For both GNU R and Python, I am using them because I want to learn them in detail. Also due to Python's minimalistic style, it makes for a good language to write code where I can post and comment. If I used something like visual basic there would be way too many lines of code for buttons and windows that it would annoy readers. So really, I am going to apply the right tools to the job.

Since this blog covers topics related to operations management, topics will include risk, quality, project management, decision theory, optimization, simulation, and process management. At a future date I plan to apply these topics to issues with computer security. At this time I will focus on issues in their traditional scope. I look forward to your posts.