Re: [R] Fourier Row and spectral analysis
Hi, i have result from other program. Original serie have 225 observing, but program use only 224. the most domain frequencies are f1=1/224 f2=1/122 f3=1/74,66 f4=1/56 f5=1/24,88 When i know these frequencies a can do, new variable ,nubmers of variavle is 2*nubmer of frequencies: t-1:225 c1-cos(2*pi*f1*t) s1-sin(2*pi*f1*t) . . . s5-sin(2*pi*f5*t) and now a can do lm(y~mean+c1+s1+c2+s2+...+s5) . Dieter Menne wrote: Peterko lanikpeter at gmail.com writes: Hello, i need to find in time serie, k=1,2,3...(how if possible) most domain frequencies and than smooth original serie by Fourier row y=mean + a1*Cos(2*pi/freq*t)+b1*Sin(2*pi/freq*t)+a2.. numbers a1,b1 ... i will have from simple regresion, i need only to find k msot domain frequensies a than i will have a1,b1,ak,bk - coeficients. It is very difficult to understand what you want, mainly the part about the simple regression in Fourier context. For simple filtering/smoothing, check http://markmail.org/message/qtyxkwlesy75 Dieter __ 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. -- View this message in context: http://www.nabble.com/Fourier-Row-and-spectral-analysis-tp22151652p22157212.html Sent from the R help mailing list archive at Nabble.com. __ 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] Fourier Row and spectral analysis
Peterko lanikpeter at gmail.com writes: Original serie have 225 observing, but program use only 224. the most domain frequencies are f1=1/224 f2=1/122 f3=1/74,66 f4=1/56 f5=1/24,88 When i know these frequencies a can do, new variable ,nubmers of variavle is 2*nubmer of frequencies: t-1:225 c1-cos(2*pi*f1*t) s1-sin(2*pi*f1*t) . . . s5-sin(2*pi*f5*t) and now a can do lm(y~mean+c1+s1+c2+s2+...+s5) Good you got it technically working, but having such highly correlated coefs like 1/224 and 1/122 in lm is asking for trouble. These are almost DC components in engineering speak for such short periods. Dieter __ 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] Fourier Row and spectral analysis
Dieter Menne wrote: Peterko lanikpeter at gmail.com writes: Original serie have 225 observing, but program use only 224. the most domain frequencies are f1=1/224 f2=1/122 f3=1/74,66 f4=1/56 f5=1/24,88 When i know these frequencies a can do, new variable ,nubmers of variavle is 2*nubmer of frequencies: t-1:225 c1-cos(2*pi*f1*t) s1-sin(2*pi*f1*t) . . . s5-sin(2*pi*f5*t) and now a can do lm(y~mean+c1+s1+c2+s2+...+s5) Good you got it technically working, but having such highly correlated coefs like 1/224 and 1/122 in lm is asking for trouble. These are almost DC components in engineering speak for such short periods. Dieter It is next problem, eliminate cerrelation, but i to not know whow to get this most domain frequencies in R, this result is from other program. __ 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. -- View this message in context: http://www.nabble.com/Fourier-Row-and-spectral-analysis-tp22151652p22160005.html Sent from the R help mailing list archive at Nabble.com. __ 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] Fourier Row and spectral analysis
Peterko lanikpeter at gmail.com writes: Hello, i need to find in time serie, k=1,2,3...(how if possible) most domain frequencies and than smooth original serie by Fourier row y=mean + a1*Cos(2*pi/freq*t)+b1*Sin(2*pi/freq*t)+a2.. numbers a1,b1 ... i will have from simple regresion, i need only to find k msot domain frequensies a than i will have a1,b1,ak,bk - coeficients. It is very difficult to understand what you want, mainly the part about the simple regression in Fourier context. For simple filtering/smoothing, check http://markmail.org/message/qtyxkwlesy75 Dieter __ 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.