Here are links to Youtubes I've done on various aspects of regression. I'll be putting up more as we progress through the technique
Simple linear regression
This is an unrealistic example just to illustrate the theory.
Multiple regression with a dummy variable
Here I show how to convert a categorical variable which is either Yes or No into 1, 0, using Excel's =(if..) function. The new variable can then be used as an independent variable in a multiple regression. Uses the 'stroke' data for predicting risk of a stroke from age, blood pressure and smoking history.
Wednesday, September 28, 2011
Monday, September 19, 2011
Project Ideas
On this page I am going to keep adding ideas I get for projects/assignments. I am not saying that you should do them or that any of them are any good. Just examples of what might be a possible start. I have seen the data in the links I posted earlier, which is how I got the idea.
1. Border crossings. Does the number of border crossings made by Canadians depend on the exchange rate? The weather? How can we predict border crossings and so optimise the staffing of the border posts?
2. Bike travel on Burrard Bridge. Data is available for bike traffic and car traffic around Hornby/Burrard. There has been a lot of controversy over the dedicated bike lanes. Does the amount of bike traffic depend on the weather (I guess if it is raining/very cold, bicyclists are possibly less likely to get into those ghastly Spandex suits?)...or perhaps the hours of daylight? or gas prices? Build a model that reflects these decisions made by bikers and give advice to City Hall. A good model could be used elsewhere for prediction.
more ideas to come....!
1. Border crossings. Does the number of border crossings made by Canadians depend on the exchange rate? The weather? How can we predict border crossings and so optimise the staffing of the border posts?
2. Bike travel on Burrard Bridge. Data is available for bike traffic and car traffic around Hornby/Burrard. There has been a lot of controversy over the dedicated bike lanes. Does the amount of bike traffic depend on the weather (I guess if it is raining/very cold, bicyclists are possibly less likely to get into those ghastly Spandex suits?)...or perhaps the hours of daylight? or gas prices? Build a model that reflects these decisions made by bikers and give advice to City Hall. A good model could be used elsewhere for prediction.
more ideas to come....!
Software Challenge!
I've been thinking about the software we are using, and the fact that some of you might have a hard time using either the library or lab workstations to access Palisades. So I have rewritten parts of the course (eg Part 3 Regression, Part 4 Time Series) so that we use basic Excel. I think we can also use Excel for Part 5 but I am not sure yet. This will make it easier for you to cooperate on your projects/assignments.
However, Excel can't do the decision-tree work we have been performing with Precision Tree, and I don't think it can Part 6 Simulation. So....
There is a lot of free open-access software about. If you can, please take a look and see if there is anything we could use. It would need to be reliable and 'safe' of course. Thanks to anyone taking on this challenge!
However, Excel can't do the decision-tree work we have been performing with Precision Tree, and I don't think it can Part 6 Simulation. So....
There is a lot of free open-access software about. If you can, please take a look and see if there is anything we could use. It would need to be reliable and 'safe' of course. Thanks to anyone taking on this challenge!
Data Sources
You can either collect your own data or find external sources, such as the Internet. Here are some sources to get you started. If you find more that you'd like to share, please let me know and I'll add them.
BC Data
Vancouver
World Bank
Kwantlen Library…business stats
http://www.kwantlen.ca/library/subjects/busi_stats.html
Weather -- including historical weather data for any city. Might be useful if you were doing the bike-riding example!
www.wunderground.com
Weather -- including historical weather data for any city. Might be useful if you were doing the bike-riding example!
www.wunderground.com
Friday, September 9, 2011
Books on reserve and textbook link
I've placed the books below on reserve in the library. These are my personal copies, so grateful for your care and attention! The first four books are source books for Part 1. The fifth book is the course text (but see the notes below)
How we decide by Jonah Lehrer
De La Mettrie’s Ghost: The Story of Decisions by Chris Nunn
Flirting with Disaster by Marc Gerstein. How some really terrible decisions have been made and why. Chapter 2 is about why humans have difficulty with probability.
Risk by Dan Gardner. Explores the ‘gut’ and ‘head’ responses we discussed in Part 1
Practical Management Science by Winston and Albright. This is the third edition. The recommended edition for the class is the fourth edition. But for our purposes this edition is fine. Note that the chapter numbering is not the same as in the course presentation, which is based on the fourth edition. I have the CD. If you need it, please let me know. But I will be posting the datasets as and when necessary.
AND one of your fellow students as found that you can check out the e-version of the course text here:
http://search.barnesandnoble.com/Data-Analysis-and-Decision-Making/S-Christian-Albright/e/9780538476126
Subscribe to:
Posts (Atom)