ashw_justin Posted June 15, 2004 Posted June 15, 2004 directions: Â -click on your name -click on "view all user's posts" -go to web address bar, change the "25&" part at the end to "1000&" and press enter (this is the "posts per window") -select all posts and copy -paste into excel -delete first 3 columns -make another column that starts at 1 from the bottom (earliest post) -make xy scatter plot from the two columns -add to cc.com gallery -post in thread to attain uber- status Quote
cracked Posted June 15, 2004 Posted June 15, 2004 Â Here we observe that unlike Dru, Cracked posts much more uniformly, with a linearity that is more precise than even the R^2 value suggests (due to the uncharacteristic lull in posts near the beginning of the data set). Conclusion: Cracked needs to go climbing. That "uncharacteristic lull" was spring break, when my Ethernet connect was pried out of my cold, dead hands. Â My pattern is more linear than Dru's because I'm not sent out to go bushwhack for a week at a time. Quote
layton Posted June 15, 2004 Posted June 15, 2004 directions: -click on your name -click on "view all user's posts" -go to web address bar, change the "25&" part at the end to "1000&" and press enter (this is the "posts per window") -select all posts and copy -paste into excel -delete first 3 columns -make another column that starts at 1 from the bottom (earliest post) -make xy scatter plot from the two columns -add to cc.com gallery -post in thread to attain uber- status  I don't have Excel or know what a xy scatter plot is. I tossed my undergrad math notes in the fire years ago. PLEASE DO MINE!!! I gave you lots o' beta and more to come when I'm in the 'Ham. Otherwise I give bad beta. Quote
layton Posted June 15, 2004 Posted June 15, 2004 Schweeet! Â Ok, for the South Twins, there is a gate and you HAVE to get permission to enter. The phone # is in the red beckey book. If they are logging they won't let you in or give you the key. BRING A BIKE! It's a maze, take mostly right turns that go uphill to a knoll/cleared spot. The lake is in the trees down and left. I'm positive the roads have changed, so it's anybody's guess. Â So someone please interpret my graph. Quote
fern Posted June 15, 2004 Posted June 15, 2004 So someone please interpret my graph. Â the steepest parts of the graph correspond to periods of time when you were looking for a date. Quote
layton Posted June 15, 2004 Posted June 15, 2004 I had a girlfriend for 50% of that graph though. Quote
layton Posted June 15, 2004 Posted June 15, 2004 oh wait! that only from 11/03. That's great! My quarter breaks are the flat parts and it gets steepest towards final exams. Makes sense. Math is neat. Quote
jjd Posted June 15, 2004 Posted June 15, 2004 We have some problems here: Â We need to see standard errors for the coefficients, t-stats, and F-values to test for overall significance. Â There is also omitted variable bias (there needs to be a dummy variable for weekends and holidays). Quote
ashw_justin Posted June 15, 2004 Posted June 15, 2004 Yeah and it's too bad that excel can't do sine/cosine wave fits. Quote
jjd Posted June 15, 2004 Posted June 15, 2004 Yeah and it's too bad that excel can't do sine/cosine wave fits. Â Maybe someone with the Matlab curve fitting tool will be kind enough to import the data and do it. Quote
ashw_justin Posted June 17, 2004 Posted June 17, 2004 Yeah and it's too bad that excel can't do sine/cosine wave fits. Â Maybe someone with the Matlab curve fitting tool will be kind enough to import the data and do it. Â Cracked, you heard the man. Get on it, colij boy. My nerd connections evaporated when I graduated. Quote
cracked Posted June 17, 2004 Posted June 17, 2004 Yeah and it's too bad that excel can't do sine/cosine wave fits. Â Maybe someone with the Matlab curve fitting tool will be kind enough to import the data and do it. Â Cracked, you heard the man. Get on it, colij boy. My nerd connections evaporated when I graduated. I've got access to Matlab, but I'm not familiar enough with matlab to do it. Sorry. Quote
ashw_justin Posted June 17, 2004 Posted June 17, 2004 That's it, you lose your d20 priviledges. Quote
jjd Posted June 17, 2004 Posted June 17, 2004 If someone gets the data into a "posts per day" format, we can run multiple regressions on Shazam. Hell, we could include dummy variables for weekends and holidays, experiment with various functional forms, and determine which model has the best fit (Shazam allows one to test for model significance, and even calculates the F-statistic for you!) Â As far as using Matlab, I have the student version which doesn't include the Curve Fitting Toolbox. Â One could fit least squares polynomials in Matlab by solving A^T*Ax = A^T*b where A is an m x n matrix of m explanatatory variables of degree n , A^T is the transpose of A, b is an n-vector of y values, and x is an m-vector of coefficients (the least squares coefficients). Â Take it from me though, Shazam is the way to go. http://shazam.ucdavis.edu Quote
Gary_Yngve Posted June 17, 2004 Posted June 17, 2004 Using a 6th-degree polynomial to model that curve is lame. The fit is snug because it's overfit, and there's absolutely no justification for why the high-degree poly model is appropriate. Quote
ashw_justin Posted June 17, 2004 Posted June 17, 2004 Using a 6th-degree polynomial to model that curve is lame. The fit is snug because it's overfit, and there's absolutely no justification for why the high-degree poly model is appropriate. Â Who said nerds don't have a sense of humor? Quote
jjd Posted June 17, 2004 Posted June 17, 2004 Using a 6th-degree polynomial to model that curve is lame. The fit is snug because it's overfit, and there's absolutely no justification for why the high-degree poly model is appropriate. Â Prove it. Quote
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.