tag:blogger.com,1999:blog-8236580214478682937.post6180138904485474543..comments2024-03-08T01:09:13.900-08:00Comments on Another Data Mining Blog: Pump Up The VolumeSali Malihttp://www.blogger.com/profile/14741877332990090234noreply@blogger.comBlogger7125tag:blogger.com,1999:blog-8236580214478682937.post-77139863367391583662012-02-01T13:53:07.386-08:002012-02-01T13:53:07.386-08:00I believe I do have all the libraries. I am trying...I believe I do have all the libraries. I am trying on Windows though(don't know if that makes a difference, but there are some libraries that are not available on windows)karmic.menacehttps://www.blogger.com/profile/02954453577302554309noreply@blogger.comtag:blogger.com,1999:blog-8236580214478682937.post-51795268823697088732012-02-01T12:53:54.701-08:002012-02-01T12:53:54.701-08:00Not seen this. I suspect though one of the librari...Not seen this. I suspect though one of the libraries is not loaded.<br /><br />library(rlecuyer)<br /><br />probably?Sali Malihttps://www.blogger.com/profile/14741877332990090234noreply@blogger.comtag:blogger.com,1999:blog-8236580214478682937.post-18323979167475437902012-02-01T11:25:52.892-08:002012-02-01T11:25:52.892-08:00SaliMali - Great blog! I was trying the parallel R...SaliMali - Great blog! I was trying the parallel Random Forest code, but it gave me an error <br /><br />"Error in checkForRemoteErrors(val) : 8 nodes produced errors; first error: object '.lec.Random.seed.table' not found<br /><br />Not sure if you have seen this before. Thought I would check.karmic.menacehttps://www.blogger.com/profile/02954453577302554309noreply@blogger.comtag:blogger.com,1999:blog-8236580214478682937.post-61744725775943682392011-07-20T13:57:31.800-07:002011-07-20T13:57:31.800-07:00@Chris - that is what I recall the guy in the shop...@Chris - that is what I recall the guy in the shop also telling me. Messing with the BIOS sounds scary though, but I might give it a try.<br />A cluster sounds fun just to try for the experience, although I suspect my new pc will be doing 95% of the work and my other 7 (one of the still with Windows95!) not a lot.<br />The prize structure is interesting, especialy as it is not clear how much exact detail you have to give a way to claim the milestone prizes. It would be possible for positions 2&3 to still 'merge' but not tell anybody and become 1&2 and take out the top 2 prizes and split the cash - something that wasn't an issue in Netflix as there was only one prize. I guess that would upset no. 1 a bit.Sali Malihttps://www.blogger.com/profile/14741877332990090234noreply@blogger.comtag:blogger.com,1999:blog-8236580214478682937.post-18709716505808068162011-07-20T13:14:11.560-07:002011-07-20T13:14:11.560-07:00Thanks for the interesting post! Seems like a nic...Thanks for the interesting post! Seems like a nice rig you've assembled. A few things:<br /><br />1. I see 8 "cores" on your Task Manager -- but with the i7, I think that means you have 4 physical cores & hyper-threading enabled. Sometimes hyper-threading helps, and sometimes (counter-intuitively) it hurts; it's kind of application specific. In any case, you might try disabling hyper-threading in the BIOS to see if that speeds things up even faster (& change threads to 4, instead of 8). <br /><br />2. Maybe you should hook up all your old PC's & create a cluster!<br /><br />3. Finally, the prize structure of the HHP is interesting -- since there's a 1st & 2nd place prize, there's actually a DIS-incentive for the top two teams to merge or cooperate to win (i.e. if position 1 & 2 merged & won & split the 1st-place prize, they'd make less than if they remained unmerged). It'll be interesting to see how the competition plays out.Chris Hefelehttps://www.blogger.com/profile/08576510162671348042noreply@blogger.comtag:blogger.com,1999:blog-8236580214478682937.post-82807689853975566842011-07-20T11:12:33.308-07:002011-07-20T11:12:33.308-07:00IMHO, you should take the money. If you trust your...IMHO, you should take the money. If you trust yourself on being able to have an edge over the others, imagine your edge plus 30.000 dollars in computing power.<br />May be I would have some doubts in the last milestone, where the money jumps from 60k to 3 million, but probably the tricks you use now want be that valuable in two years.<br />Another exception would be if you are really crushing the rest, for example, you algorithm is making 0.4300 against others doing 0.4500, you probably have some serious information you don't want to give away.<br />Very interesting blog and congrats for your current third place.<br />Cheers!Pablo Ruggiahttps://www.blogger.com/profile/04609717114174196832noreply@blogger.comtag:blogger.com,1999:blog-8236580214478682937.post-73784348161824710592011-07-20T07:48:59.721-07:002011-07-20T07:48:59.721-07:00Thanks for the post. I recently needed to get a ha...Thanks for the post. I recently needed to get a hard drive from an iMac and I was surprised how jam-packed were things inside. <br /><br />I'm using Python on a dual-core 4GB RAM PC and I'm not planning to upgrade anytime soon, since Python can only use one CPU; I'm using the other core for surfing the web and watching videos.<br /><br />Best Wishes!Sargishttps://www.blogger.com/profile/01730048467387998912noreply@blogger.com