Hi, I found the problem, now it is working fine. cheers Martin
On Thursday 04 September 2003 16:20, Martin Wegmann wrote: > Hello R user, > > I have several data frames with >100 columns and I did a linear regression > over time of each column > > df1.lm <- lapply(df1, function(x) lm(x~year)$coeff[2]) > > that worked fine and I get slope of each column oder time - until I divided > df1 by df2 > > df3 <- df1/df2 > > > df3.lm <- lapply(df3, function(x) lm(x~year)$coeff[2]) > > Error in lm.fit(x, y, offset = offset, ...) : > 0 (non-NA) cases > > df3 has cases: > X106 X107 X108 > 1 -2.200986 -2.128744 -2.126991 > 2 -2.201284 -2.179806 -1.998352 > 3 -2.201589 -2.051754 -1.918321 > 4 -2.207428 -2.024579 -2.160275 > 5 -2.088381 -2.084716 -2.033241 > 6 -2.313741 -1.905484 -2.024190 > 7 -2.232551 -2.118113 -2.123781 > 8 -2.096430 -1.787569 -2.309956 > 9 -2.029564 -1.650830 -2.038038 > 10 -1.805616 -1.936357 -1.827615 > 11 -2.427711 -2.260115 -2.192925 > 12 -2.255148 -1.624925 -2.075030 > 13 -2.040811 -1.928457 -1.985597 > 14 -2.131254 -2.126999 -2.079338 > 15 -2.407123 -2.193653 -2.162101 > 16 -2.024426 -2.168195 -2.078530 > 17 -2.224164 -1.853840 -2.150593 > 18 -2.067553 -2.140541 -1.907311 > 19 -2.142151 -2.176615 -2.015018 > 20 -2.125215 -2.051265 -1.848539 > > I did it with df1 and df2, both worked, the only difference I recognized so > far were the negative values and the amount of decimal places, therefore I > added 10 and multiplied it with 1000000 but both approaches didn't work. > > any idea what I might have done wrong? > > (using R 1.7.0 Linux) > > thanks, Martin > > ______________________________________________ > [EMAIL PROTECTED] mailing list > https://www.stat.math.ethz.ch/mailman/listinfo/r-help ______________________________________________ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help
