Re: [R] Understanding cenros Error
On 07/11/2012 05:54 AM, Rich Shepard wrote: Before reading water chemistry into a data frame I removed all missing data. Yet when I try to run cenros() to summarize a specific chemical I get an error that I do not understand: with( subset(chem, param=='Ag'), cenros(quant,ceneq1) ) Error in lm.fit(x, y, offset = offset, singular.ok = singular.ok, ...) : NA/NaN/Inf in 'y' I would like to learn what I did incorrectly so I can avoid these errors in the future. The data frame structure is str(chem) 'data.frame': 120309 obs. of 8 variables: $ site : Factor w/ 65 levels ;Influent,D-1,..: 2 2 2 2 2 2 ... $ sampdate: Date, format: 2007-12-12 2007-12-12 ... $ preeq0 : logi TRUE TRUE TRUE TRUE TRUE TRUE ... $ param : Factor w/ 37 levels Ag,Al,Alk_tot,..: 1 2 8 17 3 9 ... $ quant : num 0 0.106 1 231 231 0.011 0 0.002 0 100 ... $ ceneq1 : logi FALSE FALSE TRUE FALSE FALSE FALSE ... $ floor : num 0 0.106 0 231 231 0.011 0 0 0 100 ... $ ceiling : Factor w/ 3909 levels 0.000,0.000),..: 1 116 841 1771 ... I ran dput() on the data frame but cannot make sense of the output (a 5.5M ASCII text file). Hi Rich, I don't have the NADA package, but I suspect that the cenros function is doing something like dividing by zero. With that much data, it may be hard to pinpoint where this is occurring. I would cut my data in half, run it, cut the remainder in half, run it and so on until the error goes away. With any luck, the last slice of data that was removed won't be too large to work out which value is causing the problem and what the problem is. Jim __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] Understanding cenros Error
On Wed, 11 Jul 2012, Jim Lemon wrote: I don't have the NADA package, but I suspect that the cenros function is doing something like dividing by zero. With that much data, it may be hard to pinpoint where this is occurring. I would cut my data in half, run it, cut the remainder in half, run it and so on until the error goes away. With any luck, the last slice of data that was removed won't be too large to work out which value is causing the problem and what the problem is. Jim, That is a great idea. I'll chunk the data into smaller files and see if I can isolate the problem. Also, there are 37 different chemicals in the data set and I first should try them all and see if the problem is consistent; I tried only silver and aluminum before asking for guidance. Another potential approach that might be worth pursuing is to subset the data frame by chemical and working separately with each. As I'm just beginning to learn how to analyze left-censored data I've a long way to go. Thanks for the suggestion, Rich __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] Understanding cenros Error
An my easy but not very useful answer is that this particular subset probably violates some assumption of the cenros() model. I myself would start with simple inspections of the data, such as with( subset(chem, param=='Ag'), table(ceneq1) ) with( subset(chem, param=='Ag'), qqnorm(quant) ) with( subset(chem, param=='Ag'), range(quant) ) in the hopes that something pops out. Do you have any zeros in quant? (see ?cenros) -Don -- Don MacQueen Lawrence Livermore National Laboratory 7000 East Ave., L-627 Livermore, CA 94550 925-423-1062 On 7/10/12 12:54 PM, Rich Shepard rshep...@appl-ecosys.com wrote: Before reading water chemistry into a data frame I removed all missing data. Yet when I try to run cenros() to summarize a specific chemical I get an error that I do not understand: with( subset(chem, param=='Ag'), cenros(quant,ceneq1) ) Error in lm.fit(x, y, offset = offset, singular.ok = singular.ok, ...) : NA/NaN/Inf in 'y' I would like to learn what I did incorrectly so I can avoid these errors in the future. The data frame structure is str(chem) 'data.frame': 120309 obs. of 8 variables: $ site: Factor w/ 65 levels ;Influent,D-1,..: 2 2 2 2 2 2 ... $ sampdate: Date, format: 2007-12-12 2007-12-12 ... $ preeq0 : logi TRUE TRUE TRUE TRUE TRUE TRUE ... $ param : Factor w/ 37 levels Ag,Al,Alk_tot,..: 1 2 8 17 3 9 ... $ quant : num 0 0.106 1 231 231 0.011 0 0.002 0 100 ... $ ceneq1 : logi FALSE FALSE TRUE FALSE FALSE FALSE ... $ floor : num 0 0.106 0 231 231 0.011 0 0 0 100 ... $ ceiling : Factor w/ 3909 levels 0.000,0.000),..: 1 116 841 1771 ... I ran dput() on the data frame but cannot make sense of the output (a 5.5M ASCII text file). Pointers appreciated. Rich __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code. __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] Understanding cenros Error
On Wed, 11 Jul 2012, MacQueen, Don wrote: An my easy but not very useful answer is that this particular subset probably violates some assumption of the cenros() model. I myself would start with simple inspections of the data, such as with( subset(chem, param=='Ag'), table(ceneq1) ) with( subset(chem, param=='Ag'), qqnorm(quant) ) with( subset(chem, param=='Ag'), range(quant) ) Don, Good points. I'll do these. Do you have any zeros in quant? (see ?cenros) Many, many. My client contact is trying to determine whether they are true zeros or represent missing data. Data starts in 1994 and sometimes there are blanks (definitely missing data) and sometimes zeros ... even on the same spreadsheet row. Sigh. So we need to figure out just what they represent; perhaps the original analytical lab reports can be found and examined. I don't recall reading in Dennis' book about too many zeros so I'll check the R help. There are also errors trying cenboxplot() on the data frame. Only 9 of 37 chemicals produce results with cenros(), including 2 with censored values and 1 with 80% of concentrations censored. The remaining 28 all have censored values and/or many zeros. When I have more information, presumably positive results, I'll post here so the thread can be complete for posterity. Thanks again, Rich __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Understanding cenros Error
Before reading water chemistry into a data frame I removed all missing data. Yet when I try to run cenros() to summarize a specific chemical I get an error that I do not understand: with( subset(chem, param=='Ag'), cenros(quant,ceneq1) ) Error in lm.fit(x, y, offset = offset, singular.ok = singular.ok, ...) : NA/NaN/Inf in 'y' I would like to learn what I did incorrectly so I can avoid these errors in the future. The data frame structure is str(chem) 'data.frame': 120309 obs. of 8 variables: $ site: Factor w/ 65 levels ;Influent,D-1,..: 2 2 2 2 2 2 ... $ sampdate: Date, format: 2007-12-12 2007-12-12 ... $ preeq0 : logi TRUE TRUE TRUE TRUE TRUE TRUE ... $ param : Factor w/ 37 levels Ag,Al,Alk_tot,..: 1 2 8 17 3 9 ... $ quant : num 0 0.106 1 231 231 0.011 0 0.002 0 100 ... $ ceneq1 : logi FALSE FALSE TRUE FALSE FALSE FALSE ... $ floor : num 0 0.106 0 231 231 0.011 0 0 0 100 ... $ ceiling : Factor w/ 3909 levels 0.000,0.000),..: 1 116 841 1771 ... I ran dput() on the data frame but cannot make sense of the output (a 5.5M ASCII text file). Pointers appreciated. Rich __ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.