You could use mvrnorm once for each cluster, with a different mean vector each time (if you mean clusters in that sense.) You could draw the number of observations in a cluster randomly if that's important. Reid Huntsinger -----Original Message----- From: ge yreyt [mailto:tothri2000 at yahoo.ca] Sent: Friday, May 09, 2003 5:02 PM To: r-help Subject: [R] generate correlated dataset Hi,I want to generate a dataset, which have more than one clusters (say 2) and the elements in each cluster have high correlation (say 0.85) andelements among different clusters have low (say 0.1) or zero correlation.The correlation structure of final dataset should have a block-diagonal structure, that likes 0.85 0.85 .. 0.85 0 0 0 0 ...0 0 0...0 0 0 .......................................................0.85 0.85 .. 0.85 0 0 0 0...0 0 0 ...0 0 0 0 0 0 0 000 0000 0.85 0.85 0.85 .... 0.85.........................................................0 0 0 0 0...... 0 0 0.85 0.85 .............0.85 for each block (cluster), I can use mvrnorm command to generate data,but the key problem is I do not know how to generate data which have 0correlation when elements are among different clusters???? Thank you very much for your help!!! pin --------------------------------- Post your free ad now! Yahoo! Canada Personals [[alternate HTML version deleted]] ______________________________________________ R-help at stat.math.ethz.ch mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help ------------------------------------------------------------------------------ Notice: This e-mail message, together with any attachments, cont... {{dropped}}
generate correlated dataset
2 messages · Huntsinger, Reid, Spencer Graves
To get a target correlation between observations in different clusters, consider the following: var(Y) = E(var(Y|X))+var(E(Y|X)). The correlation between observations in a give cluster is determined by E(var(Y|X)). To get a target correction between clusters, select E(Y|X)) so var(Y) is what you want. hope this helps. spencer graves
Huntsinger, Reid wrote:
You could use mvrnorm once for each cluster, with a different mean vector each time (if you mean clusters in that sense.) You could draw the number of observations in a cluster randomly if that's important. Reid Huntsinger -----Original Message----- From: ge yreyt [mailto:tothri2000 at yahoo.ca] Sent: Friday, May 09, 2003 5:02 PM To: r-help Subject: [R] generate correlated dataset Hi,I want to generate a dataset, which have more than one clusters (say 2) and the elements in each cluster have high correlation (say 0.85) andelements among different clusters have low (say 0.1) or zero correlation.The correlation structure of final dataset should have a block-diagonal structure, that likes 0.85 0.85 .. 0.85 0 0 0 0 ...0 0 0...0 0 0 .......................................................0.85 0.85 .. 0.85 0 0 0 0...0 0 0 ...0 0 0 0 0 0 0 000 0000 0.85 0.85 0.85 .... 0.85.........................................................0 0 0 0 0...... 0 0 0.85 0.85 .............0.85 for each block (cluster), I can use mvrnorm command to generate data,but the key problem is I do not know how to generate data which have 0correlation when elements are among different clusters???? Thank you very much for your help!!! pin --------------------------------- Post your free ad now! Yahoo! Canada Personals [[alternate HTML version deleted]]
______________________________________________ R-help at stat.math.ethz.ch mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help ------------------------------------------------------------------------------ Notice: This e-mail message, together with any attachments, cont... {{dropped}} ______________________________________________ R-help at stat.math.ethz.ch mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help