MultiVoxelPatternAnalysis - Methods
location: MultiVoxelPatternAnalysis

Multi Voxel Pattern Analysis

Playing around with some methods for MVPA

Reading

See MvpaPapers (need Cambridge permissions)

Software

We're trying pymvpa

Which seems an excellent use for parallel computing with Ipython - see ParallelNotes

Covariance matrix shrinkage

Here is a set of multivariate data relating to paper production that I am using as a test bed for various things. It came from http://lib.stat.cmu.edu/datasets/papir

##############################################################################
#
# April 14. 1999       
#
# This file contains two multivariate regression data sets from 
# paper industry. They have been described and analysed in
#
# Aldrin, M. (1996), "Moderate projection pursuit regression for 
# multivariate response data", Computational Statistics and Data Analysis, 
# 21, p. 501-531.
#
# A short description is given above each data set. A more detailed 
# description is found in Aldrin (1996).
#
# These data sets are delivered by
# Magne Aldrin,
# Norwegian Computing Center, P.O. Box 114 Blindern,N-0314 Oslo
# email: magne.aldrin@nr.no
# WWW:  http://www.nr.no/~aldrin/
#
# Permission is hereby given to Statlib to redistribute these data sets. 
# They can be freely distributed and used for non-commercial purposes.
#
##############################################################################



###############################################################################
# Data set from an experiment at the paper plant Norske Skog, Skogn, Norway.
# It were described and analysed in Aldrin (1996), and published there.
#
# It consists of 30 observations (rows) and 22 variables (colums), but all 
# response variables are missing for the 28th observation.
#
# Columns 1 to 13 are response variables that describes various qualities 
# of the paper.
#
# Columns 14 to 22 are 9 predictor variables. The first three predictor 
# variables (x1 in column 14, x2 in column 15 and x3 in column 16) were 
# varied systematically through the experiment, taking the values 
# 1, 0 and -1. The next three predictor variables (columns 17 to 19) 
# are constructed as x1**2, x2**2 and x3**2. 
# The last three predictor variables (columns 20 to 22) are constructed 
# as x1*x2, x1*x3 and x2*x3. 
##############################################################################

   33.8792   30.0246    8.6891    9.6493   15.2094   11.0984   11.5204   18.5718   22.7880   17.7415   21.9271   88.4072   24.8111    1    0   -1    1    0    1    0   -1    0
   33.8792   30.2804    8.3624    9.5133   15.2094   12.2083   10.6563   20.2506   25.1815   17.2866   21.9271   87.1239   23.3873    1    0   -1    1    0    1    0   -1    0
   32.5419   31.7637    7.0558    7.8825   15.2094   12.9482    9.8643   16.8930   24.0346   15.0121   21.5845   90.5461   23.2093    0    1   -1    0    1    1    0    0   -1
   32.5419   31.3545    7.4478    8.3582   15.5262   11.4684   11.3764   18.7816   24.4335   16.8317   21.4866   91.2590   24.2060    0    1   -1    0    1    1    0    0   -1
   32.5419   31.5080    7.5785    8.4941   15.2094   11.8383   10.8723   19.2013   23.0872   17.7415   21.4376   91.5442   23.9924    0    0    0    0    0    0    0    0    0
   30.7588   32.8378    5.5532    6.1157   13.9419   11.4684   10.1523   19.3062   22.2395   18.6514   21.4376   91.1164   23.7788    0    0    0    0    0    0    0    0    0
   31.2046   33.7585    4.7039    6.3875   12.9913   14.0580    7.7042   19.2013   20.4943   20.0161   21.9271   91.6868   23.7432   -1    0    1    1    0    1    0   -1    0
   31.6503   32.3775    7.0558    7.3389   13.3082   12.2083    9.2163   17.9422   21.4915   17.7415   23.1507   90.9739   22.6397   -1    0    1    1    0    1    0   -1    0
   32.0961   32.0706    7.2518    7.6107   13.9419   12.9482    9.1443   18.7816   22.2894   17.2866   22.7591   90.1183   22.0346   -1   -1    0    1    1    0    1    0    0
   32.5419   31.7126    7.9705    8.2223   13.6251   12.9482    8.8563   18.6767   21.2422   19.1063   22.4655   89.6905   22.4261   -1   -1    0    1    1    0    1    0    0
   32.0961   32.0706    7.8398    7.4068   12.9913   12.2083    8.8563   17.1028   22.0400   16.8317   22.7102   90.8313   25.9858    0   -1    1    0    1    1    0    0   -1
   32.5419   31.4568    7.4478    7.8145   12.9913   11.4684    9.7203   18.6767   22.9376   17.7415   22.7591   90.4035   25.9858    0   -1    1    0    1    1    0    0   -1
   32.9877   31.1499    7.5131    8.4261   15.2094   11.8383   10.8003   18.9915   22.5387   18.1964   22.6613   88.9776   23.9212    0    0    0    0    0    0    0    0    0
   32.9877   31.5080    7.6438    8.1543   15.2094   12.5782   10.2963   17.2077   22.3392   16.8317   22.6613   88.9776   22.9601    0    0    0    0    0    0    0    0    0
   32.9877   31.4057    6.9905    7.9504   14.5757   11.0984   10.9443   17.1028   21.3918   17.2866   23.0039   88.6924   23.1737    1   -1    0    1    1    0   -1    0    0
   32.5419   31.3545    7.1211    7.8825   14.2588   11.0984   10.8723   15.7388   20.9929   16.3768   23.0039   89.2627   23.9212    1   -1    0    1    1    0   -1    0    0
   34.7708   29.1551    8.2318    9.0377   16.4768   12.2083   11.2324   17.3127   22.6883   16.3768   21.0461   90.1183   23.4940    1    0    1    1    0    1    0    1    0
   34.3250   29.6666    8.2971    9.1736   16.4768   13.3181   10.3683   17.2077   23.6856   15.9219   20.9971   89.8331   24.2416    1    0    1    1    0    1    0    1    0
   34.3250   29.9223    8.5584    9.4454   15.2094   13.6881    9.3603   18.2570   23.6357   16.8317   20.9971   90.5461   24.0636    0   -1   -1    0    1    1    0    0    1
   33.8792   30.0246    8.4931    9.4454   17.1106   12.5782   11.3044   18.8865   22.5387   18.1964   20.4098   90.9739   24.2772    0   -1   -1    0    1    1    0    0    1
   32.5419   31.6103    7.1211    7.8825   14.2588   13.6881    8.6403   17.4176   21.7409   17.2866   20.1161   89.8331   22.9245    0    0    0    0    0    0    0    0    0
   32.5419   31.4568    6.7945    7.8825   14.5757   13.3181    9.2163   18.8865   22.5387   18.1964   20.1161   90.5461   23.7076    0    0    0    0    0    0    0    0    0
   31.6503   32.5821    6.3372    7.1350   14.2588   13.6881    8.6403   18.4668   22.4390   17.7415   20.0672   90.2609   24.2772   -1    1    0    1    1    0   -1    0    0
   31.2046   32.6844    6.4678    7.0671   14.2588   13.3181    9.1443   20.0407   22.6883   19.1063   20.3609   89.9757   23.1737   -1    1    0    1    1    0   -1    0    0
   32.0961   31.7637    7.9051    8.4261   14.5757   14.0580    8.9283   18.0471   23.1869   16.8317   20.2630   90.1183   21.9634   -1    0   -1    1    0    1    0    1    0
   32.5419   31.5591    7.6438    7.8145   15.5262   13.6881    9.5763   19.8309   23.4363   18.1964   20.0183   89.1202   22.9245   -1    0   -1    1    0    1    0    1    0
   31.2046   32.3775    5.5532    5.9798   14.2588   10.7285   10.9443   18.1521   23.2867   16.8317   20.9971   90.9739   24.6331    0    1    1    0    1    1    0    0    1
   NA        NA         NA        NA       NA        NA        NA        NA        NA        NA        NA        NA           NA      0    1    1    0    1    1    0    0    1
   32.5419   31.5591    5.8798    7.2709   14.5757   11.8383   10.5123   18.0471   22.0899   17.7415   20.9971   90.5461   25.4519    1    1    0    1    1    0    1    0    0
   32.0961   31.5591    6.3372    7.4748   13.9419   12.5782    9.1443   17.7324   23.0373   16.8317   20.8993   89.5479   22.7821    1    1    0    1    1    0    1    0    0


###############################################################################
# Data set from an experiment at the paper plant Saugbruksforeningen, Norway.
# It were described and analysed in Aldrin (1996), but not published then.
#
# It consists of 30 observations (rows) and 41 variables (colums).
#
# Columns 1 to 32 are response variables that describes various qualities 
# of the paper.
#
# Columns 33 to 41 are 9 predictor variables. The first three predictor 
# variables  (x1 in column 33, x2 in column 34 and x3 in column 35) were 
# varied systematically through the experiment. The next three predictor 
# variables (columns 36 to 38) are constructed as x1**2, x2**2 and x3**2. 
# The last three predictor variables (columns 39 to 41) are constructed
# as x1*x2, x1*x3 and x2*x3. 
##############################################################################

  13.571  12.402  14.136  12.332  12.073  17.608  12.256  14.323  13.851  13.637  15.153  13.122  12.361  16.906  12.126  12.019  17.256  20.557  13.957  18.191  15.190  17.712  12.099  17.371  16.923  13.227  16.760  12.498  14.536  15.621  13.139  10.298   0.190   0.984   1.382   0.036   0.969   1.911   0.187   0.262   1.361
  16.890  13.494  15.583  11.565  11.801  16.775  12.821  13.104  14.193  13.153  15.297  13.319  13.396  15.280  12.969  12.701  16.012  21.743  14.662  16.675  15.624  16.480  15.002  18.045  15.646  12.235  16.510  13.149  14.536  13.571  13.467  13.269   0.861   1.528   1.588   0.742   2.334   2.520   1.316   1.367   2.425
  14.742  13.990  16.139  14.365  12.527  17.132  13.010  14.323  15.216  14.053  17.390  14.303  16.159  17.881  14.331  12.391  17.878  24.707  15.790  19.538  16.637  18.944  14.035  19.900  18.519  14.549  17.260  14.646  15.844  15.409  15.175  14.787  -0.102   0.432   0.151   0.010   0.187   0.023  -0.044  -0.015   0.065
  14.449  13.990  16.028  13.191  13.435  15.823  12.821  14.730  15.148  14.191  16.235  13.975  14.363  16.987  14.396  13.568  18.655  22.533  17.905  19.538  17.071  18.944  13.914  20.912  17.162  15.475  17.010  13.995  16.207  15.338  14.847  14.589   0.015   0.114  -0.055   0.000   0.013   0.003   0.002  -0.001  -0.006
  15.816  14.387  15.917  13.100  13.707  17.965  13.858  15.949  13.306  12.184  13.998  11.023  13.673  16.093  13.812  11.338  17.411  22.533  16.213  18.864  15.913  18.482  13.672  18.888  15.087  14.086  17.010  13.605  13.446  15.762  12.416  13.071   0.277  -1.215   1.588   0.077   1.476   2.520  -0.337   0.440  -1.929
  14.352  13.692  17.141  13.281  13.435  17.727  14.047  15.339  14.397  12.184  14.576  12.860  14.018  17.231  12.515  12.453  18.188  24.114  15.931  19.370  17.071  18.020  14.760  19.732  16.603  12.301  16.843  11.977  15.117  13.642  12.219  14.589   0.686  -1.266   1.245   0.471   1.603   1.551  -0.868   0.854  -1.577
  15.230  14.684  16.362  14.365  15.159  18.560  14.613  16.355  13.920  13.430  15.297  13.450  14.778  18.775  14.331  12.887  18.033  23.521  16.636  19.875  17.215  17.866  13.914  19.900  15.406  12.896  16.760  12.824  14.899  13.712  13.205  13.929   0.598   0.063   0.219   0.358   0.004   0.048   0.038   0.131   0.014
  14.742  16.073  16.362  12.739  13.707  18.441  15.744  15.441  13.442  13.499  14.720  12.925  13.535  17.231  13.877  12.267  19.743  23.917  16.354  19.538  17.071  18.944  14.156  19.394  15.965  14.946  15.676  13.800  14.681  16.822  13.270  13.269   0.248   0.347   0.219   0.062   0.121   0.048   0.086   0.054   0.076
  14.742  14.387  16.362  12.829  12.799  17.965  14.330  14.425  14.943  13.845  14.648  12.729  14.639  18.044  13.488  13.320  17.411  20.952  15.508  18.696  15.479  16.942  14.277  20.406  17.482  14.020  15.426  13.800  14.245  14.490  13.073  12.477  -1.766   1.549   0.287   3.119   2.399   0.083  -2.736  -0.508   0.445
  14.840  14.784  17.364  12.107  13.707  18.322  12.916  14.526  12.896  13.914  14.070  14.762  14.156  17.312  12.321  12.515  18.188  22.336  17.059  19.875  16.926  18.636  14.639  18.551  15.566  14.086  16.676  13.475  14.972  14.560  12.876  14.325  -0.920   1.388   0.151   0.845   1.926   0.023  -1.276  -0.138   0.209
  15.523  16.073  16.251  14.004  13.162  18.084  13.387  15.644  15.080  14.122  14.359  14.828  15.537  17.475  12.840  12.577  17.411  22.929  15.508  19.538  17.071  18.020  14.156  20.237  16.524  14.880  18.010  13.149  14.972  14.631  14.913  13.731   0.920   1.124   0.424   0.845   1.264   0.180   1.034   0.390   0.477
  15.914  15.081  18.699  12.468  13.435  16.775  12.538  16.457  14.261  14.191  16.163  14.106  15.261  17.800  14.137  13.196  19.588  23.126  17.200  20.549  18.228  18.482  14.639  19.900  17.242  13.888  16.927  14.711  16.425  14.772  14.387  14.457   1.007   1.600  -0.465   1.014   2.560   0.217   1.611  -0.469  -0.744
  14.840  15.478  17.141  14.185  13.616  18.798  15.178  16.050  16.035  15.160  17.101  13.778  15.813  18.450  14.591  14.435  19.899  23.521  17.764  20.886  18.083  20.022  15.244  21.081  18.998  15.277  19.595  15.297  15.989  15.197  15.241  14.853   1.182   0.063  -1.013   1.398   0.004   1.026   0.074  -1.197  -0.064
  15.914  14.982  17.809  13.823  14.705  17.727  13.576  16.660  17.058  15.576  15.225  13.844  14.916  18.857  14.655  14.188  18.344  24.510  16.777  20.044  18.517  19.098  15.123  20.575  17.641  16.203  18.094  16.794  17.152  15.338  14.847  15.382   1.445   0.538  -1.013   2.088   0.290   1.026   0.778  -1.463  -0.545
  14.059  13.891  17.141  13.281  12.981  18.084  13.858  15.034  16.581  16.753  17.462  14.500  16.504  19.426  15.304  14.621  17.722  22.336  16.213  19.033  17.215  18.174  14.035  20.912  17.721  15.211  18.594  14.972  17.152  17.741  14.059  14.787  -1.883  -0.149  -1.286   3.545   0.022   1.655   0.281   2.422   0.192
  14.938  15.478  17.030  12.558  13.525  18.322  13.387  15.238  15.284  14.745  16.019  14.500  15.882  19.832  15.044  13.754  18.966  23.719  18.187  19.370  15.913  17.404  14.035  18.888  16.603  13.888  16.927  14.646  16.643  16.751  15.504  13.731  -0.803  -0.205  -1.150   0.644   0.042   1.322   0.164   0.923   0.235
  14.449  13.097  16.139  11.745  12.073  15.704  12.067  14.019  14.193  12.945  13.565  13.385  14.847  16.906  12.386  14.064  16.634  20.952  16.495  18.022  15.624  16.942  13.309  18.888  18.360  12.367  17.260  13.865  15.626  13.995  13.927  13.269  -1.211  -0.255   1.451   1.468   0.065   2.105   0.309  -1.757  -0.371
  13.083  13.394  15.694  11.745  12.708  16.180  14.047  14.323  14.329  13.845  14.936  12.729  13.811  16.743  14.331  13.878  16.789  21.940  15.931  18.864  16.347  15.864  12.704  18.214  16.284  12.764  17.093  13.540  15.553  14.278  14.387  13.401  -0.920   0.063   1.382   0.845   0.004   1.911  -0.058  -1.271   0.087
  14.059  15.478  15.583  12.468  12.527  17.489  14.707  15.847  14.875  14.607  16.091  13.910  14.018  18.288  13.164  14.064  17.256  22.138  14.662  17.854  14.322  17.250  13.309  19.226  16.364  14.285  16.760  12.889  15.044  15.762  13.402  13.203   1.445   0.432   1.314   2.088   0.187   1.726   0.625   1.898   0.568
  14.059  14.387  15.694  14.456  12.981  18.917  14.330  15.034  13.647  14.814  14.576  13.319  14.294  16.337  14.720  12.267  17.411  22.336  15.508  19.707  16.347  18.174  13.188  19.563  17.162  14.351  16.927  13.540  14.754  14.843  14.321  14.061   1.036  -0.205   0.014   1.074   0.042   0.000  -0.212   0.014  -0.003
  13.668  14.684  15.360  13.371  12.527  17.013  13.858  14.425  15.967  14.607  15.009  13.844  14.847  17.719  15.823  13.072  17.256  21.545  15.226  17.517  16.203  17.712  13.914  18.888  16.204  15.012  17.760  13.409  14.754  14.772  14.519  13.995  -2.000  -1.105   0.219   3.998   1.220   0.048   2.209  -0.438  -0.242
  14.352  14.188  15.583  13.462  12.618  17.132  14.707  15.136  14.261  14.814  14.936  15.287  15.261  17.637  14.591  14.497  18.344  21.940  16.636  18.696  15.913  17.096  13.672  20.069  15.486  13.822  17.177  12.498  16.789  15.055  14.256  13.137  -1.912  -1.423   0.014   3.656   2.025   0.000   2.721  -0.026  -0.019
  16.597  14.684  17.809  15.088  13.525  17.251  14.990  15.034  16.308  16.407  16.019  14.631  14.708  18.857  14.072  14.126  17.567  23.126  17.341  19.538  15.913  17.404  16.212  19.900  17.162  13.359  18.678  13.800  17.225  15.550  15.241  13.797  -0.190  -0.205   0.014   0.036   0.042   0.000   0.039  -0.003  -0.003
  14.156  16.470  17.364  13.643  14.070  19.155  13.764  15.542  15.557  15.230  15.297  15.419  15.882  18.613  15.628  13.754  18.655  22.533  15.649  18.191  15.913  17.558  14.518  19.900  18.759  15.541  19.762  14.776  16.789  15.762  15.241  15.580  -0.073  -0.043   0.014   0.005   0.002   0.000   0.003  -0.001  -0.001
  14.156  15.379  16.473  13.733  15.704  18.084  15.178  15.745  14.261  13.845  15.874  12.532  15.123  17.475  14.396  13.506  19.277  23.324  17.482  20.717  17.794  19.252  16.454  20.743  17.482  13.954  17.927  14.581  16.716  16.539  16.030  14.919   0.803  -1.635   0.014   0.644   2.674   0.000  -1.313   0.011  -0.022
  16.304  15.280  17.809  15.088  14.978  19.630  15.272  17.777  13.988  14.676  15.730  15.091  14.570  18.450  14.331  15.055  19.277  23.324  17.482  20.717  17.794  19.252  16.454  20.743  17.482  13.954  17.927  14.581  16.716  16.539  16.030  14.919   0.832  -1.423  -1.971   0.692   2.025   3.884  -1.184  -1.640   2.804
  14.742  14.982  17.920  15.178  14.978  19.868  15.555  15.847  15.626  14.814  16.885  13.975  16.366  18.044  15.823  14.683  18.033  23.324  15.931  20.212  16.781  20.022  14.881  20.069  16.603  14.549  17.343  15.297  15.408  13.783  14.059  14.127   0.161  -1.499  -0.534   0.026   2.248   0.285  -0.241  -0.086   0.800
  17.281  16.569  18.365  14.365  14.433  18.560  15.555  17.574  14.329  14.122  16.957  14.303  15.330  17.231  14.201  15.303  20.209  23.126  17.059  20.044  18.373  18.328  15.728  21.924  15.885  15.144  18.928  14.972  15.044  15.197  15.044  14.589  -0.102  -1.529  -1.423   0.010   2.338   2.026   0.156   0.145   2.177
  15.426  15.974  16.919  13.733  14.796  17.965  14.518  16.050  15.216  14.261  15.874  15.353  16.090  19.182  14.526  15.117  18.499  23.126  15.931  20.212  17.505  19.714  14.156  19.563  17.162  15.078  18.511  14.451  16.789  14.843  15.110  13.599   0.394   0.806  -1.423   0.155   0.650   2.026   0.318  -0.561  -1.147
  15.914  15.676  17.030  14.275  14.615  18.798  14.141  15.441  15.762  13.845  15.947  15.419  16.435  17.475  15.044  13.692  18.655  22.533  15.790  18.528  17.071  18.020  14.035  18.551  16.204  14.086  18.261  14.321  15.044  14.843  14.650  14.259  -0.219   1.124  -1.355   0.048   1.264   1.836  -0.246   0.297  -1.523

None: MultiVoxelPatternAnalysis (last edited 2013-03-08 10:28:26 by localhost)