| Re: clustering [message #54986 is a reply to message #54895] |
Thu, 19 July 2007 13:46   |
Vince Hradil
Messages: 574
Registered: December 1999
|
Senior Member |
|
|
On Jul 19, 11:12 am, nivedita.raghun...@gmail.com wrote:
> Here is a subset of my data.
>
> IDL> help,pos1
> POS1 FLOAT = Array[7, 53]
>
> IDL> print,pos1
> 0.910300 0.413400 -0.0221000 0.00300000 -150.250
> 129.510 -13.0400
> 0.910200 0.413400 -0.0223000 0.00370000 -150.280
> 129.460 -13.0200
> 0.910200 0.413500 -0.0228000 0.00360000 -150.300
> 129.400 -13.1300
> 0.910200 0.413400 -0.0231000 0.00310000 -150.190
> 129.520 -13.0700
> 0.910200 0.413400 -0.0226000 0.00320000 -150.220
> 129.580 -13.0800
> 0.910200 0.413600 -0.0224000 0.00460000 -150.510
> 129.040 -13.1000
> 0.910200 0.413500 -0.0221000 0.00250000 -150.210
> 129.560 -13.0000
> 0.910200 0.413500 -0.0223000 0.00340000 -150.310
> 129.420 -13.1000
> 0.910200 0.413500 -0.0225000 0.00350000 -150.160
> 129.620 -13.0900
> 0.910200 0.413500 -0.0224000 0.00240000 -150.090
> 129.720 -13.0100
> 0.930600 0.365500 -0.0216000 0.00170000 -147.800
> 125.760 -16.7500
> 0.930500 0.365600 -0.0220000 0.000900000 -147.650
> 125.160 -16.6800
> 0.930500 0.365700 -0.0222000 0.00230000 -147.930
> 125.370 -16.8100
> 0.930500 0.365700 -0.0217000 0.00280000 -148.090
> 125.750 -16.8600
> 0.930400 0.365800 -0.0225000 0.00240000 -147.800
> 125.400 -16.8200
> 0.930400 0.365800 -0.0213000 0.00430000 -148.490
> 124.950 -16.7800
> 0.930400 0.365800 -0.0220000 0.00210000 -147.910
> 126.000 -16.7200
> 0.930400 0.365800 -0.0220000 0.00200000 -147.830
> 125.560 -16.6900
> 0.930400 0.365900 -0.0216000 0.00250000 -148.080
> 125.490 -16.7700
> 0.930400 0.365800 -0.0224000 0.00230000 -147.870
> 125.980 -16.6200
> 0.897600 0.439600 -0.0331000 0.00790000 -147.060
> 130.970 -6.02000
> 0.897600 0.439500 -0.0334000 0.00720000 -146.790
> 130.520 -6.13000
> 0.897500 0.439600 -0.0337000 0.00770000 -146.820
> 130.660 -6.13000
> 0.897500 0.439600 -0.0328000 0.00750000 -147.160
> 130.790 -6.13000
> 0.897600 0.439600 -0.0331000 0.00680000 -146.860
> 130.570 -6.07000
> 0.897600 0.439600 -0.0335000 0.00700000 -146.830
> 130.660 -6.12000
> 0.897600 0.439500 -0.0326000 0.00750000 -147.090
> 130.870 -6.08000
> 0.897600 0.439600 -0.0327000 0.00750000 -146.880
> 130.610 -6.14000
> 0.897600 0.439500 -0.0336000 0.00810000 -146.980
> 130.560 -6.25000
> 0.897600 0.439500 -0.0331000 0.00800000 -147.130
> 130.820 -6.19000
> 0.897500 0.439600 -0.0332000 0.00800000 -147.000
> 130.600 -6.25000
> 0.871700 0.488800 -0.0332000 0.0102000 -146.260
> 133.480 -1.14000
> 0.871600 0.488900 -0.0330000 0.0111000 -146.390
> 133.540 -1.29000
> 0.871600 0.488900 -0.0347000 0.00920000 -145.690
> 132.630 -1.26000
> 0.871700 0.488800 -0.0337000 0.0103000 -146.100
> 133.330 -1.44000
> 0.871700 0.488700 -0.0336000 0.0104000 -146.310
> 133.610 -1.58000
> 0.871700 0.488800 -0.0340000 0.00950000 -145.820
> 132.840 -1.33000
> 0.872000 0.488200 -0.0335000 0.00960000 -146.040
> 133.140 -1.95000
> 0.872000 0.488200 -0.0330000 0.00820000 -145.910
> 133.210 -1.83000
> 0.872000 0.488300 -0.0333000 0.0100000 -146.040
> 133.110 -1.82000
> 0.872100 0.488200 -0.0330000 0.00880000 -146.000
> 133.150 -1.83000
> 0.872000 0.488200 -0.0335000 0.00900000 -145.910
> 133.210 -1.85000
> 0.873000 0.487300 -0.0227000 0.000700000 -143.720
> 132.260 -6.08000
> 0.872900 0.487300 -0.0230000 0.000100000 -143.630
> 132.350 -6.07000
> 0.872900 0.487300 -0.0235000 0.000500000 -143.560
> 132.370 -6.14000
> 0.872900 0.487300 -0.0234000 -0.000300000 -143.430
> 132.520 -6.15000
> 0.872900 0.487300 -0.0231000 0.000700000 -143.670
> 132.280 -6.15000
> 0.872900 0.487300 -0.0237000 0.000200000 -143.480
> 132.430 -6.07000
> 0.872900 0.487300 -0.0231000 0.000500000 -143.550
> 132.450 -6.03000
> 0.872900 0.487300 -0.0241000 -0.000200000 -143.440
> 132.450 -6.11000
> 0.872900 0.487300 -0.0237000 0.000400000 -143.470
> 132.490 -6.05000
> 0.873000 0.487300 -0.0228000 0.000600000 -143.700
> 132.270 -6.03000
> 0.872900 0.487300 -0.0235000 -0.000200000 -143.430
> 132.450 -6.10000
>
> IDL> weights=clust_wts(pos1,n_clusters=5)
> IDL> print,weights
> 0.159265 0.119451 0.113155 0.180601 0.0680267
> 0.243488 0.116014
> 0.874568 0.483835 -0.0256644 0.00231699 -144.219
> 132.388 -5.40240
> 0.113501 0.127323 0.0985566 0.247231 0.225678
> 0.0779656 0.109745
> 0.238006 0.236222 0.127174 0.0261984 0.266028
> 0.0180832 0.0882878
> 0.0301962 0.232814 0.209770 0.146116 0.235975
> 0.134589 0.0105386
>
> IDL> result=cluster(pos1,weights,n_clusters=5)
> IDL> print,result(uniq(result))
> 1
>
> The 5 clusters are pretty distinct but "cluster" does a hopeless job
> identifying them. I tried scaling the data but that too finds 3
> clusters in the end. Any ideas?
>
> On Jul 18, 3:07 pm, Conor <cmanc...@gmail.com> wrote:
>
>> On Jul 17, 11:36 am, nivedita.raghun...@gmail.com wrote:
>
>>> Hi all,
>
>>> I am trying to cluster a (7 x n) array with n_clusters=5. Visually I
>>> can see 5 distinct clusters, but when I do clust_wts the cluster
>>> centroids don't end up right. No matter what options I give, clust_wts
>>> refuses to find the clusters.
>
>>> Any idea on whats going on ?
>
>>> Thanks,
>>> Nivedita
>
>> No idea. More information might be helpful. It's quite possible
>> though that the the clust_wts algorithm just doesn't work for your
>> particular data set, at least not as well as you apparently want it to.
The scales of all the variables are very different. Looks like
everything is in the second cluster, due to the large (abs value) of
the 5th dimension.
You might want to standardize() your data first
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
