The R Project SVN
R-packages
(root)/trunk/cluster/tests/clara-NAs.Rout.save – Rev 8309
Rev 8273 | Blame | Compare with Previous | Last modification | View Log | Download | RSS feed
R Under development (unstable) (2023-11-29 r85646) -- "Unsuffered Consequences"Copyright (C) 2023 The R Foundation for Statistical ComputingPlatform: x86_64-pc-linux-gnuR is free software and comes with ABSOLUTELY NO WARRANTY.You are welcome to redistribute it under certain conditions.Type 'license()' or 'licence()' for distribution details.R is a collaborative project with many contributors.Type 'contributors()' for more information and'citation()' on how to cite R or R packages in publications.Type 'demo()' for some demos, 'help()' for on-line help, or'help.start()' for an HTML browser interface to help.Type 'q()' to quit R.> library(cluster)>> x <- cbind(c(0, -4, -22, -14, 0, NA, -28, 1, 10, -1,+ 100 + c(13, 0, 2, 4, 7, 8, 1)),+ c(-5, -14, NA, -35, -30, NA, 7, 2, -18, 13,+ 47, 64, 48, NA, NA, 44, 65))> x[,1] [,2][1,] 0 -5[2,] -4 -14[3,] -22 NA[4,] -14 -35[5,] 0 -30[6,] NA NA[7,] -28 7[8,] 1 2[9,] 10 -18[10,] -1 13[11,] 113 47[12,] 100 64[13,] 102 48[14,] 104 NA[15,] 107 NA[16,] 108 44[17,] 101 65> (d <- dist(x,'manhattan'))1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 162 133 44 364 44 31 165 25 20 44 196 NA NA NA NA NA7 40 45 12 56 65 NA8 8 21 46 52 33 NA 349 23 18 64 41 22 NA 63 2910 19 30 42 61 44 NA 33 13 4211 165 178 270 209 190 NA 181 157 168 14812 169 182 244 213 194 NA 185 161 172 152 3013 155 168 248 199 180 NA 171 147 158 138 12 1814 208 216 252 236 208 NA 264 206 188 210 18 8 415 214 222 258 242 214 NA 270 212 194 216 12 14 10 616 157 170 260 201 182 NA 173 149 160 140 8 28 10 8 217 171 184 246 215 196 NA 187 163 174 154 30 2 18 6 12 28> summary(d, na.rm = TRUE) # max = 270Min. 1st Qu. Median Mean 3rd Qu. Max. NA's2.00 27.25 147.50 114.55 188.50 270.00 16> ## First call with "trace" (seg.fault typically later ...):> try( clara(x, k=2, metric="manhattan", sampsize=10, trace = 3) )C clara(): (nsam,nran,n) = (10,5,17); 'large_sample',- clara sample 1 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 1[ntt=7, nunfs=0] .. nsel[1:10]= 6 7 8 9 10 12 13 14 16 17 -> dysta2() gave dyst_toomany_NA --> new sample.- clara sample 2 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 3[ntt=7, nunfs=1] .. nsel[1:10]= 1 4 7 9 11 12 13 15 16 17 -> dysta2();clara -> s:= max{dys[1..45]} = 270;bswap2(*, s=270), 1. BUILD: new repr. 7new repr. 1after build: medoids are 1 7and min.dist dysma[1:n] are0 44 40 23 12 18 0 10 10 18--> sky = sum_j D_j= 175swp new 8 <-> 7 old; decreasing diss. by -18Last swap: new 8 <-> 7 old; decreasing diss. by 1end{bswap2}: sky = 157selec() -> 'NAfs' obj= 7.41176- clara sample 3 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 6[ntt=7, nunfs=2] .. nsel[1:10]= 1 2 3 5 8 9 10 11 13 17 -> dysta2();clara -> s:= max{dys[1..45]} = 270;bswap2(*, s=270), 1. BUILD: new repr. 5new repr. 9after build: medoids are 5 9and min.dist dysma[1:n] are8 21 46 33 0 29 13 12 0 18--> sky = sum_j D_j= 180swp new 1 <-> 5 old; decreasing diss. by -18Last swap: new 1 <-> 5 old; decreasing diss. by 1end{bswap2}: sky = 162selec() -> 'NAfs' obj= 7.41176- clara sample 4 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 5[ntt=7, nunfs=3] .. nsel[1:10]= 1 2 3 4 7 8 9 10 13 14 -> dysta2();clara -> s:= max{dys[1..45]} = 264;bswap2(*, s=264), 1. BUILD: new repr. 1new repr. 10after build: medoids are 1 10and min.dist dysma[1:n] are0 13 44 44 40 8 23 19 4 0--> sky = sum_j D_j= 195Last swap: new 9 <-> 10 old; decreasing diss. by 0end{bswap2}: sky = 195selec() -> 'NAfs' obj= 7.41176- clara sample 5 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 16[ntt=7, nunfs=4] .. nsel[1:10]= 2 3 4 6 7 8 9 10 11 17 -> dysta2() gave dyst_toomany_NA --> new sample.C clara() -> best sample _found_ ; nbest[1:10] =c(0,0,0,0,0,0,0,0,0,0)Error in clara(x, k = 2, metric = "manhattan", sampsize = 10, trace = 3) :Observation 6 has *only* NAs --> omit it for clusteringIn addition: Warning message:In clara(x, k = 2, metric = "manhattan", sampsize = 10, trace = 3) :Distance computations with NAs: using correct instead of pre-2016 wrong formula.Use 'correct.d=FALSE' to get previous results or set 'correct.d=TRUE' explicitlyto suppress this warning.> ## Originally:already shows the problem: nbest[] = c(0,0,...,0) must be WRONG!!> ## Now: gives the proper error message.>> ## S-plus 6.1.2 (rel.2 for Linux, 2002) gives> ##> cc <- clara(x, k=2, metric="manhattan", samples=2, sampsize=10)> ## Problem in .Fortran("clara",: Internal error: data for decrementing> ## ref.count didn't point to a valid arena (0x0), while calling subroutine clara>> ## The large example from clara.R -- made small enough to still provoke> ## the "** dysta2() ... OUT" problem {no longer!}> x <- matrix(c(0, 3, -4, 62, 1, 3, -7, 45, 36, 46, 45, 54, -10,+ 51, 49, -5, 13, -6, 49, 52, 57, 39, -1, 55, 68, -3, 51, 11, NA,+ 9, -3, 50, NA, 58, 9, 52, 12, NA, 47, -12, -6, -9, 5, 30, 38,+ 54, -5, 39, 50, 50, 54, 43, 7, 64, 55, 4, 0, 72, 54, 37, 59,+ -1, 8, 43, 50, -2, 56, -8, 43, 6, 4, 48, -2, 14, 45, 49, 56,+ 51, 45, 11, 10, 42, 50, 2, -12, 3, 1, 2, 2, -14, -4, 8, 0, 3,+ -11, 8, 5, 14, -1, 9, 0, 19, 10, -2, -9, 9, 2, 16, 10, 4, 1,+ 12, 7, -4, 27, -8, -9, -9, 2, 8, NA, 13, -23, -3, -5, 1, 15,+ -3, 5, -9, -5, 14, 8, 7, -4, 26, 20, 10, 8, 17, 4, 14, 23, -2,+ 23, 2, 16, 5, 5, -3, 12, 5, 14, -2, 4, 2, -2, 7, 9, 1, -15, -1,+ 9, 23, 1, 7, 13, 2, -11, 16, 12, -11, -14, 2, 6, -8),+ ncol = 2)> str(x) # 88 x 2num [1:88, 1:2] 0 3 -4 62 1 3 -7 45 36 46 ...> try(clara(x, 2, samples = 20, trace = 3))# 2nd sample did show dysta2() problemC clara(): (nsam,nran,n) = (44,20,88);- clara sample 1 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 2[ntt=44, nunfs=0] .. nsel[1:44]= 2 3 4 6 9 10 12 14 15 18 19 20 24 25 26 28 31 35 38 42 47 48 51 53 54 57 60 61 64 66 68 70 71 73 74 75 76 77 78 79 80 81 82 88 -> dysta2();clara -> s:= max{dys[1..946]} = 78.6448;bswap2(*, s=78.6448), 1. BUILD: new repr. 19new repr. 9after build: medoids are 9 19and min.dist dysma[1:n] are21.2 7.07 9.9 2.83 5.66 5 5.1 9.22 0 11.31.41 6.71 6.32 8.49 7.07 12.7 1.41 33.9 0 14.17.07 18.9 5.39 4.24 15.5 31.1 5.66 5.66 5.66 4.241.41 8.49 11.3 22.6 2.83 4.12 13 0 3.61 51.41 17 9.22 12.7--> sky = sum_j D_j= 385.677Last swap: new 6 <-> 9 old; decreasing diss. by 0.939294end{bswap2}: sky = 385.677selec() -> 'NAfs' obj= 2.59347- clara sample 2 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 12[ntt=44, nunfs=1] .. nsel[1:44]= 1 2 3 5 6 7 9 12 17 19 26 27 28 29 30 38 39 42 43 45 47 50 52 54 55 56 58 59 60 61 62 64 67 68 71 74 75 76 77 79 80 81 83 84 -> dysta2();clara -> s:= max{dys[1..946]} = 81.7435;bswap2(*, s=81.7435), 1. BUILD: new repr. 16new repr. 43after build: medoids are 16 43and min.dist dysma[1:n] are1.41 21.2 7.07 1.41 2.83 17 5.66 5 14.1 1.417.07 15 12.7 14.1 14.1 0 4.24 14.1 8.49 9.97.07 2 8.6 14.1 12.1 4.24 1.41 5.66 5.66 5.665.66 5.66 4.24 1.41 11.3 2.83 5.83 11 0 5.11.41 17 0 17--> sky = sum_j D_j= 331.98Last swap: new 17 <-> 43 old; decreasing diss. by 0.492109end{bswap2}: sky = 331.98selec() -> 'NAfs' obj= 2.55701- clara sample 3 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 14[ntt=44, nunfs=2] .. nsel[1:44]= 1 3 4 6 7 9 12 14 15 16 18 19 20 29 30 31 32 36 38 39 40 44 46 47 48 49 51 53 54 56 57 60 62 64 65 66 67 73 75 77 79 81 82 87 -> dysta2();clara -> s:= max{dys[1..946]} = 77.8781;bswap2(*, s=77.8781), 1. BUILD: new repr. 19new repr. 35after build: medoids are 19 35and min.dist dysma[1:n] are1.41 7.07 9.9 2.83 17 5.66 6.4 5.1 4.12 4.2411.3 1.41 2.83 14.1 14.1 1.41 6 5.66 0 3.165.66 18.4 8.06 7.07 16.3 6 7.21 4.24 14 4.2431.1 5.66 5.66 5.66 0 4.24 4.24 22.6 7.07 05.1 17 8.25 7.07--> sky = sum_j D_j= 338.587end{bswap2}: sky = 338.587selec() -> 'NAfs' obj= 2.57726- clara sample 4 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 70[ntt=44, nunfs=3] .. nsel[1:44]= 1 3 8 9 14 15 16 17 19 20 22 23 28 30 31 32 34 35 36 37 38 39 40 41 45 46 47 49 54 56 57 65 66 67 69 70 71 74 76 77 78 84 86 88 -> dysta2();clara -> s:= max{dys[1..946]} = 77.8781;bswap2(*, s=77.8781), 1. BUILD: new repr. 21new repr. 32after build: medoids are 21 32and min.dist dysma[1:n] are1.41 7.07 7.81 5.66 5.1 4.12 4.24 14.1 1.41 2.834.24 0 12.7 14.1 1.41 6 8.06 33.9 5.66 8.490 3.16 5.66 5.66 9.9 8.06 7.07 6 14 4.2431.1 0 4.24 4.24 4.24 8.49 11.3 2.83 9.06 07.07 17 1.41 12.7--> sky = sum_j D_j= 325.933end{bswap2}: sky = 325.933selec() -> 'NAfs' obj= 2.57726- clara sample 5 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 80[ntt=44, nunfs=4] .. nsel[1:44]= 1 2 3 5 7 8 11 13 14 20 22 23 26 28 30 31 33 34 37 38 39 41 45 46 47 50 51 52 57 59 61 64 67 71 76 77 79 80 81 82 85 86 87 88 -> dysta2() gave dyst_toomany_NA --> new sample.- clara sample 6 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 5[ntt=44, nunfs=5] .. nsel[1:44]= 2 3 4 5 6 8 10 12 19 20 21 23 24 25 29 30 31 32 33 37 39 41 42 45 46 48 50 53 54 59 61 66 68 69 71 72 73 79 80 82 84 85 86 87 -> dysta2() gave dyst_toomany_NA --> new sample.- clara sample 7 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 17[ntt=44, nunfs=6] .. nsel[1:44]= 2 3 6 7 8 12 14 16 17 18 20 22 26 27 29 30 31 32 33 35 36 37 42 44 45 46 49 52 54 58 59 61 62 63 65 67 70 74 75 77 78 79 87 88 -> dysta2() gave dyst_toomany_NA --> new sample.- clara sample 8 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 67[ntt=44, nunfs=7] .. nsel[1:44]= 2 3 6 7 8 9 11 13 14 18 30 31 32 34 35 37 38 40 43 44 47 48 49 52 54 55 56 58 59 60 66 67 68 70 71 72 73 75 80 83 84 85 87 88 -> dysta2();clara -> s:= max{dys[1..946]} = 85.5102;bswap2(*, s=85.5102), 1. BUILD: new repr. 17new repr. 9after build: medoids are 9 17and min.dist dysma[1:n] are21.2 7.07 2.83 17 9.9 5.66 2.83 1.41 0 11.314.1 1.41 9.9 9.22 33.9 8.49 0 5.66 8.49 18.47.07 13.9 1.41 8.25 13.9 5.66 4.24 1.41 4.24 5.664.24 4.24 1.41 8.49 11.3 0 22.6 11.3 1.41 7.0717 21.2 7.07 12.7--> sky = sum_j D_j= 384.698end{bswap2}: sky = 384.698selec() -> 'NAfs' obj= 2.73432- clara sample 9 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 67[ntt=44, nunfs=8] .. nsel[1:44]= 2 4 6 7 8 11 12 13 14 15 17 19 20 21 24 27 29 30 31 33 34 35 36 39 42 45 46 48 50 51 52 54 55 58 60 61 62 65 67 78 80 84 86 88 -> dysta2() gave dyst_toomany_NA --> new sample.- clara sample 10 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 5[ntt=44, nunfs=9] .. nsel[1:44]= 1 3 5 6 7 9 10 14 15 16 17 18 19 20 21 23 28 29 30 32 33 36 37 39 40 44 46 47 51 53 54 55 56 57 65 69 70 74 76 81 82 84 86 87 -> dysta2() gave dyst_toomany_NA --> new sample.- clara sample 11 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 66[ntt=44, nunfs=10] .. nsel[1:44]= 1 3 4 5 6 11 13 14 15 18 19 21 28 30 31 32 33 34 39 40 41 42 43 46 47 57 58 59 63 65 66 67 71 72 73 74 75 78 79 80 83 84 87 88 -> dysta2() gave dyst_toomany_NA --> new sample.- clara sample 12 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 21[ntt=44, nunfs=11] .. nsel[1:44]= 4 5 6 8 9 10 13 14 15 16 17 21 23 25 27 28 30 35 36 41 44 46 47 49 50 54 55 56 57 59 61 62 64 65 66 68 71 72 74 75 76 81 83 84 -> dysta2();clara -> s:= max{dys[1..946]} = 78.3135;bswap2(*, s=78.3135), 1. BUILD: new repr. 5new repr. 2after build: medoids are 2 5and min.dist dysma[1:n] are26.2 0 3.61 9.49 0 13.5 11 20.5 13.9 6.3215 21.6 2.24 32.1 26.6 12.8 12 24.4 17.9 8.610.8 18.1 7.21 20.5 14.9 29.4 26.2 3.61 23 21.123 3.61 7 16.6 3.61 9.22 9.49 12.6 13 9.8522.2 14.2 15.7 11--> sky = sum_j D_j= 623.732swp new 43 <-> 5 old; decreasing diss. by -205.358Last swap: new 43 <-> 5 old; decreasing diss. by 1end{bswap2}: sky = 418.375selec() -> 'NAfs' obj= 3.17257- clara sample 13 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 4[ntt=44, nunfs=12] .. nsel[1:44]= 3 4 5 14 15 16 17 19 20 21 24 25 26 29 30 31 34 35 38 40 41 43 47 49 50 52 55 57 58 60 61 63 64 65 66 68 72 73 74 79 81 83 86 88 -> dysta2();clara -> s:= max{dys[1..946]} = 84.1487;bswap2(*, s=84.1487), 1. BUILD: new repr. 19new repr. 34after build: medoids are 19 34and min.dist dysma[1:n] are7.07 9.9 1.41 5.1 4.12 4.24 14.1 1.41 2.83 8.065.39 8.49 7.07 14.1 14.1 1.41 8.06 33.9 0 5.665.66 8.49 7.07 6 4 7.62 10.3 31.1 1.41 5.665.66 15.6 5.66 0 4.24 1.41 0 22.6 2.83 5.117 2 1.41 12.7--> sky = sum_j D_j= 340.1end{bswap2}: sky = 340.1selec() -> 'NAfs' obj= 2.57726- clara sample 14 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 78[ntt=44, nunfs=13] .. nsel[1:44]= 5 7 8 9 10 11 12 18 19 21 23 27 28 29 31 32 33 35 36 38 39 46 50 51 52 56 57 58 59 60 64 65 66 68 72 73 75 77 78 80 84 86 87 88 -> dysta2() gave dyst_toomany_NA --> new sample.- clara sample 15 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 73[ntt=44, nunfs=14] .. nsel[1:44]= 2 3 8 10 18 25 26 27 29 31 33 34 35 41 42 43 44 46 47 48 49 53 54 56 57 58 59 60 63 69 70 71 72 73 75 76 77 79 81 84 85 86 87 88 -> dysta2() gave dyst_toomany_NA --> new sample.- clara sample 16 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 43[ntt=44, nunfs=15] .. nsel[1:44]= 5 6 7 8 9 12 14 16 18 22 23 24 27 28 29 30 34 35 36 39 40 41 43 45 56 57 59 60 62 64 65 67 69 70 71 73 74 75 79 81 83 85 86 87 -> dysta2();clara -> s:= max{dys[1..946]} = 75.1665;bswap2(*, s=75.1665), 1. BUILD: new repr. 15new repr. 41after build: medoids are 15 41and min.dist dysma[1:n] are12.7 17 2.83 6.4 15.7 5 7.07 9.9 25.5 13.614.1 5 15 1.41 0 0 8.06 19.8 8.49 4.248.49 19.8 5.66 12.6 9.9 45.3 5.66 14.8 8.49 9.92 10 9.9 22.6 25.5 8.49 11.3 5.83 5.1 2.830 7.07 15.6 21.2--> sky = sum_j D_j= 479.721end{bswap2}: sky = 479.721selec() -> 'NAfs' obj= 3.31146- clara sample 17 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 22[ntt=44, nunfs=16] .. nsel[1:44]= 4 6 9 10 11 12 13 16 19 22 26 27 29 30 33 34 37 38 39 42 43 48 51 54 55 57 60 61 62 63 64 66 69 72 73 75 76 77 78 81 82 85 86 87 -> dysta2() gave dyst_toomany_NA --> new sample.- clara sample 18 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 39[ntt=44, nunfs=17] .. nsel[1:44]= 1 4 8 13 15 19 20 23 25 26 27 28 30 31 33 34 36 37 39 41 42 43 44 45 46 47 50 54 55 57 59 60 62 64 65 66 67 72 73 78 79 81 82 85 -> dysta2() gave dyst_toomany_NA --> new sample.- clara sample 19 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 72[ntt=44, nunfs=18] .. nsel[1:44]= 1 4 5 6 10 12 13 14 17 18 19 22 23 25 27 30 31 32 33 38 39 40 41 42 44 45 46 48 55 57 58 59 60 61 66 67 69 70 72 74 83 84 85 86 -> dysta2() gave dyst_toomany_NA --> new sample.- clara sample 20 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 48[ntt=44, nunfs=19] .. nsel[1:44]= 1 3 4 5 7 10 11 13 14 20 22 23 31 32 33 34 35 36 37 40 41 42 43 44 48 50 52 53 55 56 62 63 68 71 72 73 74 75 81 82 83 84 86 88 -> dysta2() gave dyst_toomany_NA --> new sample.C clara() -> best sample _found_ ; nbest[1:44] =c(0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0)Error in clara(x, 2, samples = 20, trace = 3) :Observation 33 has *only* NAs --> omit it for clusteringIn addition: Warning message:In clara(x, 2, samples = 20, trace = 3) :Distance computations with NAs: using correct instead of pre-2016 wrong formula.Use 'correct.d=FALSE' to get previous results or set 'correct.d=TRUE' explicitlyto suppress this warning.> ## To see error message for > 1 missing:> try(clara(rbind(NA,x), 2))Error in clara(rbind(NA, x), 2) :Observations 1,34 have *only* NAs --> omit them for clustering!In addition: Warning message:In clara(rbind(NA, x), 2) :Distance computations with NAs: using correct instead of pre-2016 wrong formula.Use 'correct.d=FALSE' to get previous results or set 'correct.d=TRUE' explicitlyto suppress this warning.>> x <- x[-33,]> ## still had the ** dysta2() .. OUT" problem {no longer!}> c2 <- clara(x, 2, samples = 12, trace = 3)C clara(): (nsam,nran,n) = (44,12,87); 'large_sample',- clara sample 1 finding 1st... new k{ran}:.. kall: FALSE,... nrx [0:1]= 0 0... nsel[0:0]= 2[ntt=43, nunfs=0] .. nsel[1:44]= 1 7 8 11 14 16 17 21 22 26 29 30 32 33 34 36 37 39 40 41 43 44 45 46 48 49 51 52 54 55 56 58 62 64 66 68 69 71 74 82 83 84 85 86 -> dysta2();clara -> s:= max{dys[1..946]} = 76.5376;bswap2(*, s=76.5376), 1. BUILD: new repr. 17new repr. 40after build: medoids are 17 40and min.dist dysma[1:n] are1.41 17 6.4 2.83 7.07 4.24 14.1 7.28 4.24 7.0714.1 14.1 4 8.06 33.9 8.49 0 5.66 5.66 14.118.4 9.9 6.4 7.07 8 2 8.6 4.24 12.1 4.2431.1 5.66 15.6 2 4.24 4.24 8.49 0 5.83 017 21.2 1.41 7.07--> sky = sum_j D_j= 384.62end{bswap2}: sky = 384.621st proper sample obj= 7.92464- clara sample 2 finding 1st... new k{ran}:.. kall: T,... nrx [0:1]= 37 82... nsel[0:0]= 70[ntt=43, nunfs=0] .. nsel[1:44]= 4 8 10 11 13 15 16 18 20 21 22 23 24 25 31 32 33 34 35 36 37 40 41 43 45 46 48 50 52 55 62 64 65 68 69 71 72 77 81 82 84 85 86 87 -> dysta2();clara -> s:= max{dys[1..946]} = 82.7103;bswap2(*, s=82.7103), 1. BUILD: new repr. 21new repr. 40after build: medoids are 21 40and min.dist dysma[1:n] are9.9 6.4 4.47 2.83 1.41 2.24 4.24 11.3 4.47 7.284.24 0 5 8.49 1.41 4 8.06 33.9 5.66 8.490 5.66 14.1 18.4 6.4 7.07 8 5.66 4.24 4.2415.6 2 4.24 4.24 8.49 0 22.6 5.1 8.94 021.2 1.41 7.07 12.7--> sky = sum_j D_j= 321.274end{bswap2}: sky = 321.274obj= 7.92464- clara sample 3 finding 1st... new k{ran}:.. kall: T,... nrx [0:1]= 37 82... nsel[0:0]= 38[ntt=43, nunfs=0] .. nsel[1:44]= 2 5 10 11 14 18 21 22 23 24 25 26 27 29 33 34 36 37 39 41 42 43 45 47 49 50 55 57 58 60 62 64 67 68 71 73 75 77 79 82 83 84 85 87 -> dysta2();clara -> s:= max{dys[1..946]} = 85.5102;bswap2(*, s=85.5102), 1. BUILD: new repr. 18new repr. 32after build: medoids are 18 32and min.dist dysma[1:n] are21.2 1.41 4 2.83 5.1 11.3 8.06 4.24 0 5.398.49 7.07 13 14.1 8.06 33.9 8.49 0 5.66 14.18.49 18.4 8.06 16.3 4 7.21 4.24 1.41 4.47 5.6615.6 0 1.41 4.24 0 2.83 9.06 7.07 1.41 217 21.2 1.41 12.7--> sky = sum_j D_j= 350.699Last swap: new 40 <-> 32 old; decreasing diss. by 0.7867end{bswap2}: sky = 350.699new best obj= 7.91099- clara sample 4 finding 1st... new k{ran}:.. kall: T,... nrx [0:1]= 37 64... nsel[0:0]= 85[ntt=43, nunfs=0] .. nsel[1:44]= 2 4 5 6 8 10 11 12 15 18 21 24 25 26 28 31 33 37 40 42 43 47 49 50 52 54 55 57 58 59 60 61 62 63 64 71 72 73 74 78 79 81 82 86 -> dysta2();clara -> s:= max{dys[1..946]} = 78.3135;bswap2(*, s=78.3135), 1. BUILD: new repr. 18new repr. 43after build: medoids are 18 43and min.dist dysma[1:n] are21.2 9.9 1.41 2.83 6.4 4.47 2.83 5 2.24 11.37.28 5 8.49 7.07 12.7 1.41 8.06 0 5.66 8.4918.4 17.8 2 5.66 4.24 12.1 4.24 1.41 5.66 5.665.66 5.66 15.6 5.66 2 0 22.6 2.83 5.83 5.11.41 8.94 0 7.07--> sky = sum_j D_j= 297.276end{bswap2}: sky = 297.276obj= 7.92464- clara sample 5 finding 1st... new k{ran}:.. kall: T,... nrx [0:1]= 37 64... nsel[0:0]= 84[ntt=43, nunfs=0] .. nsel[1:44]= 4 6 9 10 12 13 15 16 17 18 20 21 24 26 27 31 34 35 36 37 40 41 42 43 45 51 52 53 54 55 56 62 64 65 67 68 69 71 72 73 74 77 79 82 -> dysta2();clara -> s:= max{dys[1..946]} = 76.5376;bswap2(*, s=76.5376), 1. BUILD: new repr. 20new repr. 44after build: medoids are 20 44and min.dist dysma[1:n] are9.9 2.83 5.66 4.47 5 1.41 2.24 4.24 14.1 11.34.47 7.28 5 7.07 15 1.41 33.9 5.66 8.49 05.66 14.1 8.49 18.4 6.4 8.6 4.24 14.1 12.1 4.2431.1 15.6 2 4.24 1.41 4.24 8.49 0 22.6 2.835.83 5.1 1.41 0--> sky = sum_j D_j= 350.799end{bswap2}: sky = 350.799obj= 7.92464- clara sample 6 finding 1st... new k{ran}:.. kall: T,... nrx [0:1]= 37 64... nsel[0:0]= 33[ntt=43, nunfs=0] .. nsel[1:44]= 1 5 7 9 11 13 14 15 17 18 22 26 27 28 34 35 36 37 38 43 46 47 48 50 51 54 55 56 57 61 63 64 66 69 71 73 74 75 76 77 78 80 81 82 -> dysta2();clara -> s:= max{dys[1..946]} = 82.0244;bswap2(*, s=82.0244), 1. BUILD: new repr. 18new repr. 19after build: medoids are 18 19and min.dist dysma[1:n] are1.41 1.41 17 5.66 2.83 1.41 5.66 5.39 14.1 11.34.24 7.07 12.6 12.7 33.9 5.66 8.49 0 0 177.07 13.6 5.83 9.9 4.47 11.3 4.24 31.1 1.41 5.665.66 3.16 4.24 8.49 0 2.83 6.32 8.25 0 8.492.83 17 5.1 4.24--> sky = sum_j D_j= 339.187end{bswap2}: sky = 339.187obj= 8.0873- clara sample 7 finding 1st... new k{ran}:.. kall: T,... nrx [0:1]= 37 64... nsel[0:0]= 26[ntt=43, nunfs=0] .. nsel[1:44]= 1 3 4 5 9 10 11 18 19 21 23 24 25 30 37 38 39 40 46 47 50 54 55 56 59 62 64 65 66 67 68 70 71 72 75 76 79 80 81 82 83 84 85 86 -> dysta2();clara -> s:= max{dys[1..946]} = 82.7103;bswap2(*, s=82.7103), 1. BUILD: new repr. 15new repr. 27after build: medoids are 15 27and min.dist dysma[1:n] are1.41 7.07 9.9 1.41 5.66 4 2.83 11.3 1.41 8.060 5.39 8.49 14.1 0 3.16 5.66 5.66 7.07 16.37.21 10.3 4.24 31.1 5.66 15.6 0 4.24 4.24 1.414.24 11.3 0 22.6 9.06 0 1.41 17 8.25 217 21.2 1.41 7.07--> sky = sum_j D_j= 325.427end{bswap2}: sky = 325.427obj= 7.91099- clara sample 8 finding 1st... new k{ran}:.. kall: T,... nrx [0:1]= 37 64... nsel[0:0]= 3[ntt=43, nunfs=0] .. nsel[1:44]= 4 5 6 10 12 15 16 17 19 20 21 22 23 24 25 26 27 28 29 31 37 39 41 42 44 45 46 49 50 51 56 57 61 62 63 64 68 73 75 77 78 80 81 85 -> dysta2();clara -> s:= max{dys[1..946]} = 84.1487;bswap2(*, s=84.1487), 1. BUILD: new repr. 21new repr. 36after build: medoids are 21 36and min.dist dysma[1:n] are9.9 1.41 2.83 4 6.4 4.12 4.24 14.1 1.41 2.838.06 4.24 0 5.39 8.49 7.07 13 12.7 14.1 1.410 5.66 14.1 8.49 9.9 8.06 7.07 4 7.21 7.6231.1 1.41 5.66 15.6 5.66 0 4.24 2.83 9.06 7.075.1 17 8.25 1.41--> sky = sum_j D_j= 312.333end{bswap2}: sky = 312.333obj= 7.91099- clara sample 9 finding 1st... new k{ran}:.. kall: T,... nrx [0:1]= 37 64... nsel[0:0]= 59[ntt=43, nunfs=0] .. nsel[1:44]= 1 3 5 9 11 13 17 22 23 25 29 31 32 34 37 38 40 42 43 44 47 49 53 55 56 58 62 63 64 66 68 69 70 71 72 73 74 75 76 78 81 82 84 86 -> dysta2();clara -> s:= max{dys[1..946]} = 82.7103;bswap2(*, s=82.7103), 1. BUILD: new repr. 15new repr. 29after build: medoids are 15 29and min.dist dysma[1:n] are1.41 7.07 1.41 5.66 2.83 1.41 14.1 4.24 0 8.4914.1 1.41 6 33.9 0 3.16 5.66 8.49 18.4 9.916.3 4 14 4.24 31.1 4.47 15.6 5.66 0 4.244.24 8.49 11.3 0 22.6 2.83 7.07 9.06 0 5.18.25 2 21.2 7.07--> sky = sum_j D_j= 356.571Last swap: new 16 <-> 29 old; decreasing diss. by 0.311473end{bswap2}: sky = 356.571obj= 7.91099- clara sample 10 finding 1st... new k{ran}:.. kall: T,... nrx [0:1]= 37 64... nsel[0:0]= 5[ntt=43, nunfs=0] .. nsel[1:44]= 2 4 8 11 12 13 22 24 25 26 27 29 31 34 36 37 39 41 42 44 47 48 49 50 53 57 58 59 60 61 62 63 64 66 67 70 71 72 75 77 79 82 84 87 -> dysta2();clara -> s:= max{dys[1..946]} = 85.5102;bswap2(*, s=85.5102), 1. BUILD: new repr. 16new repr. 33after build: medoids are 16 33and min.dist dysma[1:n] are21.2 9.9 7.81 2.83 6.4 1.41 4.24 5.39 8.49 7.0713 14.1 1.41 33.9 8.49 0 5.66 14.1 8.49 9.916.3 6 4 7.21 14 1.41 4.47 5.66 5.66 5.6615.6 5.66 0 4.24 1.41 11.3 0 22.6 9.06 7.071.41 2 21.2 12.7--> sky = sum_j D_j= 368.598Last swap: new 42 <-> 33 old; decreasing diss. by 0.115684end{bswap2}: sky = 368.598obj= 7.91099- clara sample 11 finding 1st... new k{ran}:.. kall: T,... nrx [0:1]= 37 64... nsel[0:0]= 30[ntt=43, nunfs=0] .. nsel[1:44]= 2 7 8 9 10 16 17 20 22 23 24 26 27 29 34 35 36 37 38 41 44 47 48 49 50 51 52 53 54 59 60 61 63 64 67 68 69 75 76 78 80 81 84 85 -> dysta2();clara -> s:= max{dys[1..946]} = 80.9938;bswap2(*, s=80.9938), 1. BUILD: new repr. 18new repr. 19after build: medoids are 18 19and min.dist dysma[1:n] are21.2 17 7.28 5.66 1.41 4.24 14.1 5.1 4.24 08.54 7.07 12.6 14.1 33.9 5.66 8.49 0 0 14.19.9 13.6 5.83 5.83 9.9 4.47 4.24 17 11.3 5.665.66 5.66 5.66 3.16 1.41 4.24 8.49 8.25 0 2.8317 5.1 21.2 1.41--> sky = sum_j D_j= 362.717end{bswap2}: sky = 362.717obj= 8.0873- clara sample 12 finding 1st... new k{ran}:.. kall: T,... nrx [0:1]= 37 64... nsel[0:0]= 44[ntt=43, nunfs=0] .. nsel[1:44]= 1 2 3 7 11 12 14 18 19 20 21 28 30 31 32 33 34 37 38 39 40 42 49 51 52 53 55 59 63 64 68 69 70 72 74 76 77 78 79 81 84 85 86 87 -> dysta2();clara -> s:= max{dys[1..946]} = 80.9938;bswap2(*, s=80.9938), 1. BUILD: new repr. 18new repr. 30after build: medoids are 18 30and min.dist dysma[1:n] are1.41 21.2 7.07 17 2.83 6.4 5.1 11.3 1.41 2.838.06 12.7 14.1 1.41 6 8.06 33.9 0 3.16 5.665.66 8.49 4 7.62 4.24 14 4.24 5.66 5.66 04.24 8.49 11.3 22.6 7.07 0 7.07 5.1 1.41 8.2521.2 1.41 7.07 12.7--> sky = sum_j D_j= 347.279end{bswap2}: sky = 347.279obj= 7.91099C clara() -> best sample _found_ ; nbest[1:44] =c(2,5,10,11,14,18,21,22,23,24,25,26,27,29,33,34,36,37,39,41,42,43,45,47,49,50,55,57,58,60,62,64,67,68,71,73,75,77,79,82,83,84,85,87)resul(), black() and return() from C.Warning message:In clara(x, 2, samples = 12, trace = 3) :Distance computations with NAs: using correct instead of pre-2016 wrong formula.Use 'correct.d=FALSE' to get previous results or set 'correct.d=TRUE' explicitlyto suppress this warning.> c2. <- clara(x, 2, samples = 12, trace = 1, correct.d=TRUE)C clara(): (nsam,nran,n) = (44,12,87); 'large_sample',- clara sample 1 [ntt=43, nunfs=0] -> dysta2(); obj= 7.92464- clara sample 2 [ntt=43, nunfs=0] -> dysta2(); obj= 7.92464- clara sample 3 [ntt=43, nunfs=0] -> dysta2(); obj= 7.91099- clara sample 4 [ntt=43, nunfs=0] -> dysta2(); obj= 7.92464- clara sample 5 [ntt=43, nunfs=0] -> dysta2(); obj= 7.92464- clara sample 6 [ntt=43, nunfs=0] -> dysta2(); obj= 8.0873- clara sample 7 [ntt=43, nunfs=0] -> dysta2(); obj= 7.91099- clara sample 8 [ntt=43, nunfs=0] -> dysta2(); obj= 7.91099- clara sample 9 [ntt=43, nunfs=0] -> dysta2(); obj= 7.91099- clara sample 10 [ntt=43, nunfs=0] -> dysta2(); obj= 7.91099- clara sample 11 [ntt=43, nunfs=0] -> dysta2(); obj= 8.0873- clara sample 12 [ntt=43, nunfs=0] -> dysta2(); obj= 7.91099C clara() -> best sample _found_ resul(), black() and return() from C.> p2g <- pam(daisy(x,"gower"), k=2, trace = 3)pam()'s bswap(*, s=0.785, pamonce=0): build 2 medoids:new repr. 37new repr. 64after build: medoids are 37 64and min.dist dysma[1:n] are0.02 0.3 0.1 0.131 0.02 0.04 0.24 0.0898 0.08 0.02380.04 0.0738 0.02 0.056 0.046 0.06 0.2 0.16 0.02 0.03190.0817 0.06 0 0.0498 0.12 0.1 0.136 0.18 0.2 0.20.02 0.06 0.0576 0.48 0.08 0.12 0 0.0279 0.08 0.080.2 0.12 0.119 0.131 0.0938 0.1 0.185 0.06 0.04 0.08380.0717 0.06 0.0833 0.12 0.06 0.388 0.02 0.0438 0.08 0.080.08 0.22 0.08 0 0.06 0.06 0.02 0.06 0.12 0.160 0.32 0.04 0.0798 0.096 0 0.076 0.0398 0.02 0.240.0676 0.02 0.24 0.3 0.02 0.1 0.18end{bswap()}, end{cstat()}> if(FALSE) { ## disabled clara(*, "gower") for now (2023-11-30):+ c2g <- clara(x, 2, samples = 12, sampsize=nrow(x), trace = 2, metric = "gower", pamLike=TRUE, correct.d=TRUE)+ (icall <- which(names(c2) == "call"))+ ## c2g and p2g are *quite* different !+ table(c2g$clustering,+ p2g$clustering)+ ## 1 2+ ## 1 40 32+ ## 2 15 0 << not *one* pair of {2,2} !?!++ stopifnot(exprs = {+ all.equal(c2[-icall], c2.[-icall])+ })+ }# no "gower" for now>> data(xclara)> suppressWarnings(RNGversion("3.5.0")) # back compatibility of results> set.seed(123)> xclara[sample(nrow(xclara), 50),] <- NA> try( clara(xclara, k = 3) ) #-> "nice" error message {.. first 12 missing obs} :Error in clara(xclara, k = 3) :50 observations (74,126,137,308,411,423,438,451,642,686,689,735 ...) have *only* NAs --> omit them for clustering!In addition: Warning message:In clara(xclara, k = 3) :Distance computations with NAs: using correct instead of pre-2016 wrong formula.Use 'correct.d=FALSE' to get previous results or set 'correct.d=TRUE' explicitlyto suppress this warning.> ## Error in clara(xclara, k = 3) :> ## 50 observations (74,126,137,308,411,423,438,451,642,686,689,735 ...) have *only* NAs> ## --> omit them for clustering!>> proc.time()user system elapsed0.149 0.039 0.205