Thanks for the note Jan. Not sure what happened, but hopefully it's back on track now.
Hello Ned. Couple of trends stopped on 16 Apr, any ideas why?
Agree with you Grant
Improve English to better understand maths: http://speakenglishfluently.net
It's a good idea. Maybe you can give it a try? The data is all out there.
A lot of the top 25 contributors have plateaued and are not active anymore. Can you make another one that has a list of the Top 10 most active contributors? You could check the top 50 or so contributors and find out which have increased their score the most in the past, say, 3 months, and display their curves.
Haha, thanks! Only one we've found so far :D
This little blue dot in the sky... Very nice!!
Very nice plot. I didn't know about Trendy until recently, and this makes it look very tempting...
More problem groups (challenges) might help reinvigorate the site.
As J.R! said... :-)
I would like to apply this wonderfull representation to time series in a clinical research (patients for players and visits for dates). Congratulations !
This is really nice... Thanks Alfonso!
Thanks! I wanted to pay tribute to everyone that made it (or will make it) into this list, so many stories in these curves!
This is gorgeous Alfonso! My compliments to the chef.
Oh, common, they closed the API only to open it again after several days?...
- New background color for ach subplot
- Degrees diplayed using Tex symbol : \circC
Nice and its realistic....watch the clustering of different moods...its amazingly correct.
Well, Google has finally closed the old search API. Now it is required to have an API key to use JSON search. I don't see a way to use it without exposing it to the wild (not that it matters a lot, but still).
That's strange, here it returns error, but when a new plot is created using the same code, it works.
You can delete directly the colected data on the Trend page (http://www.mathworks.com/matlabcentral/trendy/trends/2326/trend_data?page=2), but only if you are the owner (creator) of the Trend.
Otherwise you can exclude the sample on your plot script:
- find the sample
- delete the sample
- plot the results
this process can be automated or hard coded, as you like.
Normalised to what?
Weather.com changed their API dramatically.
Yes, that's good. How about another trendy that works with the top 10 contributors?
Sharp drops are due to stock splits. Probably should take that into account in the plotting.
I've ordered the lines in the legend based on where they are from top to bottom on the last day. That should help disambiguate a little.
Better. The two lime green lines are still pretty close by my eyes and my monitors.
This chart is so beautiful, to see the top players on Cody. One day my name will be there!
Is there any way to delete the erroneous point (Early October) for Jobs?
I tightened the time view a little and tried to make the colors more obviously unique.
There are two curves for blue, green, and red. We don't know which red curve is which person. Can you make all the colors be unique?
Thanks for the note. I've reported the problem to the right people here.
The comment about Cody was a copy-and-paste mistake. Thanks for pointing it out. Running the code outside the context of Trendy is a little tricky, since you need to bring the data down from the site. Here's a post that might help: http://blogs.mathworks.com/community/2014/07/11/weather-prediction-how-far-can-you-go/
Plot is broken. Can you fix it?
How do you run this? Copying and pasting the code gives the error
Undefined function or variable 'time1756'.
Error in test3 (line 5)
t = time1756;
The first line says "Top players in Cody" - do I need to do something in Cody first?
The plot is not broken. Trendy just thinks that it is.
Rapidly dropping in popularity!
Thanks, Ned. Very nice plot!
You are right, Jan! Congratulations, and sorry for not having your name up there right away!
It seems that player table needs an update :-)
This is a cumsum histogram, isn't it? Nice!
About your code inprovement: Yes! you're right! but the function urlfilter does also an urlreal call, and since I have taken yet the hmtl code in the loop, I decided to get the number by my self... ;-)
Yes I found not fair to say I'm the top #N from 25k Players, wenn soooo many does not have score at all... ;-)
Ha ha :-D Newer expected so many ghost-players. Like it!
As well as your code, you can use:
pattern='1 – 50 of ';
[-urlfilter([pre zero],pattern) 0]+urlfilter(pre,pattern);
They updated the function "updatetrend.m"... isn't it? :-/
Maybe is it more interesting to collect this data hourly... ;-)
How can we see the code that was used to extract the data from the web-site? I am interested in doing something similar but cannot figure out how to do the data scraping. Many of the sites do not seem to use pure html but have java mixed in. I am interested to find out more.