I N F L U E N C E V A L U E S R E P O R T order attributes by influence values = true ============================================= AutoClass CLASSIFICATION for the 214 cases in /home/tove/p/autoclass-c/data/glass/glassc.db2 /home/tove/p/autoclass-c/data/glass/glass-3c.hd2 with log-A (approximate marginal likelihood) = -11187.745 from classification results file /home/tove/p/autoclass-c/data/glass/glassc.results-bin and using models /home/tove/p/autoclass-c/data/glass/glass-mnc.model - index = 0 ORDER OF PRESENTATION: * Summary of the generating search. * Weight ordered list of the classes found & class strength heuristic. * List of class cross entropies with respect to the global class. * Ordered list of attribute influence values summed over all classes. * Class listings, ordered by class weight. _____________________________________________________________________________ _____________________________________________________________________________ SEARCH SUMMARY 4 tries over 3 seconds _______________ SUMMARY OF 10 BEST RESULTS _________________________ ## PROBABILITY exp(-10897.738) N_CLASSES 5 FOUND ON TRY 3 *SAVED* -1 PROBABILITY exp(-11187.745) N_CLASSES 7 FOUND ON TRY 4 *SAVED* -2 PROBABILITY exp(-11229.517) N_CLASSES 3 FOUND ON TRY 2 PROBABILITY exp(-11236.431) N_CLASSES 2 FOUND ON TRY 1 ## - report filenames suffix _____________________________________________________________________________ _____________________________________________________________________________ CLASSIFICATION HAS 7 POPULATED CLASSES (max global influence value = 1.225) We give below a heuristic measure of class strength: the approximate geometric mean probability for instances belonging to each class, computed from the class parameters and statistics. This approximates the contribution made, by any one instance "belonging" to the class, to the log probability of the data set w.r.t. the classification. It thus provides a heuristic measure of how strongly each class predicts "its" instances. Class Log of class Relative Class Normalized num strength class strength weight class weight 0 -4.41e+01 1.00e+00 118 0.551 1 -5.80e+01 9.75e-07 28 0.131 2 -5.06e+01 1.57e-03 17 0.079 3 -5.89e+01 3.75e-07 17 0.079 4 -6.03e+01 9.76e-08 14 0.066 5 -5.40e+01 5.36e-05 12 0.057 6 -5.16e+01 6.02e-04 8 0.037 CLASS DIVERGENCES The class divergence, or cross entropy w.r.t. the single class classification, is a measure of how strongly the class probability distribution function differs from that of the database as a whole. It is zero for identical distributions, going infinite when two discrete distributions place probability 1 on differing values of the same attribute. Class class cross entropy Class Normalized num w.r.t. global class weight class weight 0 1.10e+01 118 0.551 1 5.45e+00 28 0.131 2 8.01e+00 17 0.079 3 8.48e+00 17 0.079 4 1.03e+01 14 0.066 5 8.97e+00 12 0.057 6 9.60e+00 8 0.037 ORDERED LIST OF NORMALIZED ATTRIBUTE INFLUENCE VALUES SUMMED OVER ALL CLASSES This gives a rough heuristic measure of relative influence of each attribute in differentiating the classes from the overall data set. Note that "influence values" are only computable with respect to the model terms. When multiple attributes are modeled by a single dependent term (e.g. multi_normal_cn), the term influence value is distributed equally over the modeled attributes. num description I-*k 011 Log RI: refractive index 1.000 012 Log Na: Wt.% Sodium oxide 1.000 013 Log Mg: Wt.% Magnesium oxide 1.000 014 Log Al: Wt.% Aluminum oxide 1.000 015 Log Si: Wt.% Silicon oxide 1.000 016 Log K: Wt.% Potassium oxide 1.000 017 Log Ca: Wt.% Calcium oxide 1.000 018 Log Ba: Wt.% Barium oxide 1.000 019 Log Fe: Wt.% Iron oxide 1.000 000 Id number ----- 001 RI: refractive index ----- 002 Na: Wt.% Sodium oxide ----- 003 Mg: Wt.% Magnesium oxide ----- 004 Al: Wt.% Aluminum oxide ----- 005 Si: Wt.% Silicon oxide ----- 006 K: Wt.% Potassium oxide ----- 007 Ca: Wt.% Calcium oxide ----- 008 Ba: Wt.% Barium oxide ----- 009 Fe: Wt.% Iron oxide ----- 010 Type of glass ----- CLASS LISTINGS: These listings are ordered by class weight -- * j is the zero-based class index, * k is the zero-based attribute index, and * l is the zero-based discrete attribute instance index. Within each class, the covariant and independent model terms are ordered by their term influence value I-jk. Covariant attributes and discrete attribute instance values are both ordered by their significance value. Significance values are computed with respect to a single class classification, using the divergence from it, abs( log( Prob-jkl / Prob-*kl)), for discrete attributes and the relative separation from it, abs( Mean-jk - Mean-*k) / StDev-jk, for numerical valued attributes. For the SNcm model, the value line is followed by the probabilities that the value is known, for that class and for the single class classification. Entries are attribute type dependent, and the corresponding headers are reproduced with each class. In these -- * num/t denotes model term number, * num/a denotes attribute number, * t denotes attribute type, * mtt denotes model term type, and * I-jk denotes the term influence value for attribute k in class j. This is the cross entropy or Kullback-Leibler distance between the class and full database probability distributions (see interpretation-c.text). * Mean StDev -jk -jk The estimated mean and standard deviation for attribute k in class j. * |Mean-jk - The absolute difference between the Mean-*k|/ two means, scaled w.r.t. the class StDev-jk standard deviation, to get a measure of the distance between the attribute means in the class and full data. * Mean StDev The estimated mean and standard -*k -*k deviation for attribute k when the model is applied to the data set as a whole. * Prob-jk is known 1.00e+00 Prob-*k is known 9.98e-01 The SNcm model allows for the possibility that data values are unknown, and models this with a discrete known/unknown probability. The gaussian normal for known values is then conditional on the known probability. In this instance, we have a class where all values are known, as opposed to a database where only 99.8% of values are known. CLASS 0 - weight 118 normalized weight 0.551 relative strength 1.00e+00 ******* class cross entropy w.r.t. global class 1.10e+01 ******* Model file: /home/tove/p/autoclass-c/data/glass/glass-mnc.model - index = 0 Numbers: numb/t = model term number; numb/a = attribute number Model term types (mtt): (multi_normal_cn MNcn) REAL ATTRIBUTE (t = R) |Mean-jk - numb t mtt description I-jk Mean StDev Mean-*k|/ Mean StDev t a -jk -jk StDev-jk -*k -*k 00 13 R MNcn Log Mg: Wt.% Magnesi 1.225 ( 1.28e+00 4.35e-02) 3.15e+01 (-9.37e-02 2.60e+00) um oxide 00 18 R MNcn Log Ba: Wt.% Barium 1.225 (-5.30e+00 2.85e-02) 3.04e+01 (-4.43e+00 1.93e+00) oxide 00 16 R MNcn Log K: Wt.% Potassi 1.225 (-6.98e-01 4.70e-01) 1.65e+00 (-1.47e+00 1.74e+00) um oxide 00 17 R MNcn Log Ca: Wt.% Calcium 1.225 ( 2.14e+00 6.19e-02) 6.66e-01 ( 2.18e+00 1.45e-01) oxide 00 12 R MNcn Log Na: Wt.% Sodium 1.225 ( 2.58e+00 2.96e-02) 5.20e-01 ( 2.59e+00 6.09e-02) oxide 00 11 R MNcn Log RI: refractive i 1.225 (-6.58e-01 3.46e-03) 2.91e-01 (-6.57e-01 5.82e-03) ndex 00 14 R MNcn Log Al: Wt.% Aluminu 1.225 ( 2.44e-01 2.58e-01) 2.41e-01 ( 3.06e-01 3.70e-01) m oxide 00 19 R MNcn Log Fe: Wt.% Iron ox 1.225 (-4.07e+00 1.69e+00) 5.96e-02 (-4.18e+00 1.66e+00) ide 00 15 R MNcn Log Si: Wt.% Silicon 1.225 ( 4.29e+00 6.50e-03) 5.33e-02 ( 4.29e+00 1.07e-02) oxide Correlation matrix (row & column indices are attribute numbers) 11 12 13 14 15 16 17 18 19 11 1.000 0.431 0.464 -0.804 -0.721 -0.814 0.857 0.000 0.176 12 0.431 1.000 0.383 -0.411 -0.738 -0.525 0.171 0.000 -0.107 13 0.464 0.383 1.000 -0.393 -0.463 -0.366 0.146 0.000 0.077 14 -0.804 -0.411 -0.393 1.000 0.475 0.816 -0.777 0.000 -0.066 15 -0.721 -0.738 -0.463 0.475 1.000 0.599 -0.576 0.000 -0.136 16 -0.814 -0.525 -0.366 0.816 0.599 1.000 -0.740 0.000 -0.053 17 0.857 0.171 0.146 -0.777 -0.576 -0.740 1.000 0.000 0.210 18 0.000 0.000 0.000 0.000 0.000 0.000 0.000 1.000 0.000 19 0.176 -0.107 0.077 -0.066 -0.136 -0.053 0.210 0.000 1.000 CLASS 1 - weight 28 normalized weight 0.131 relative strength 9.75e-07 ******* class cross entropy w.r.t. global class 5.45e+00 ******* Model file: /home/tove/p/autoclass-c/data/glass/glass-mnc.model - index = 0 Numbers: numb/t = model term number; numb/a = attribute number Model term types (mtt): (multi_normal_cn MNcn) REAL ATTRIBUTE (t = R) |Mean-jk - numb t mtt description I-jk Mean StDev Mean-*k|/ Mean StDev t a -jk -jk StDev-jk -*k -*k 00 18 R MNcn Log Ba: Wt.% Barium 0.606 (-5.62e-01 1.81e+00) 2.14e+00 (-4.43e+00 1.93e+00) oxide 00 13 R MNcn Log Mg: Wt.% Magnesi 0.606 (-4.42e+00 2.28e+00) 1.89e+00 (-9.37e-02 2.60e+00) um oxide 00 12 R MNcn Log Na: Wt.% Sodium 0.606 ( 2.67e+00 5.95e-02) 1.34e+00 ( 2.59e+00 6.09e-02) oxide 00 11 R MNcn Log RI: refractive i 0.606 (-6.61e-01 3.84e-03) 1.13e+00 (-6.57e-01 5.82e-03) ndex 00 16 R MNcn Log K: Wt.% Potassi 0.606 (-3.89e+00 2.24e+00) 1.08e+00 (-1.47e+00 1.74e+00) um oxide 00 14 R MNcn Log Al: Wt.% Aluminu 0.606 ( 6.89e-01 4.13e-01) 9.29e-01 ( 3.06e-01 3.70e-01) m oxide 00 15 R MNcn Log Si: Wt.% Silicon 0.606 ( 4.29e+00 1.07e-02) 8.51e-01 ( 4.29e+00 1.07e-02) oxide 00 19 R MNcn Log Fe: Wt.% Iron ox 0.606 (-4.77e+00 1.12e+00) 5.25e-01 (-4.18e+00 1.66e+00) ide 00 17 R MNcn Log Ca: Wt.% Calcium 0.606 ( 2.11e+00 1.44e-01) 4.85e-01 ( 2.18e+00 1.45e-01) oxide Correlation matrix (row & column indices are attribute numbers) 11 12 13 14 15 16 17 18 19 11 1.000 -0.232 -0.295 0.439 -0.438 -0.047 0.305 0.532 0.050 12 -0.232 1.000 -0.210 -0.355 0.058 -0.557 -0.136 0.032 -0.021 13 -0.295 -0.210 1.000 -0.084 -0.234 0.302 -0.610 -0.134 -0.197 14 0.439 -0.355 -0.084 1.000 -0.637 0.038 0.404 0.623 0.206 15 -0.438 0.058 -0.234 -0.637 1.000 -0.190 0.155 -0.786 -0.021 16 -0.047 -0.557 0.302 0.038 -0.190 1.000 -0.452 -0.074 -0.061 17 0.305 -0.136 -0.610 0.404 0.155 -0.452 1.000 0.055 0.232 18 0.532 0.032 -0.134 0.623 -0.786 -0.074 0.055 1.000 0.106 19 0.050 -0.021 -0.197 0.206 -0.021 -0.061 0.232 0.106 1.000 CLASS 2 - weight 17 normalized weight 0.079 relative strength 1.57e-03 ******* class cross entropy w.r.t. global class 8.01e+00 ******* Model file: /home/tove/p/autoclass-c/data/glass/glass-mnc.model - index = 0 Numbers: numb/t = model term number; numb/a = attribute number Model term types (mtt): (multi_normal_cn MNcn) REAL ATTRIBUTE (t = R) |Mean-jk - numb t mtt description I-jk Mean StDev Mean-*k|/ Mean StDev t a -jk -jk StDev-jk -*k -*k 00 16 R MNcn Log K: Wt.% Potassi 0.890 (-5.67e-01 2.21e-01) 4.09e+00 (-1.47e+00 1.74e+00) um oxide 00 13 R MNcn Log Mg: Wt.% Magnesi 0.890 ( 9.05e-01 3.86e-01) 2.58e+00 (-9.37e-02 2.60e+00) um oxide 00 12 R MNcn Log Na: Wt.% Sodium 0.890 ( 2.57e+00 4.10e-02) 5.93e-01 ( 2.59e+00 6.09e-02) oxide 00 17 R MNcn Log Ca: Wt.% Calcium 0.890 ( 2.22e+00 9.01e-02) 4.30e-01 ( 2.18e+00 1.45e-01) oxide 00 19 R MNcn Log Fe: Wt.% Iron ox 0.890 (-3.33e+00 2.06e+00) 4.09e-01 (-4.18e+00 1.66e+00) ide 00 14 R MNcn Log Al: Wt.% Aluminu 0.890 ( 3.78e-01 1.87e-01) 3.83e-01 ( 3.06e-01 3.70e-01) m oxide 00 18 R MNcn Log Ba: Wt.% Barium 0.890 (-3.96e+00 1.79e+00) 2.65e-01 (-4.43e+00 1.93e+00) oxide 00 11 R MNcn Log RI: refractive i 0.890 (-6.57e-01 3.45e-03) 1.14e-01 (-6.57e-01 5.82e-03) ndex 00 15 R MNcn Log Si: Wt.% Silicon 0.890 ( 4.29e+00 6.54e-03) 7.60e-02 ( 4.29e+00 1.07e-02) oxide Correlation matrix (row & column indices are attribute numbers) 11 12 13 14 15 16 17 18 19 11 1.000 0.702 -0.779 0.267 -0.846 -0.538 0.800 0.021 0.178 12 0.702 1.000 -0.521 -0.156 -0.661 -0.702 0.563 -0.071 0.270 13 -0.779 -0.521 1.000 -0.470 0.728 0.562 -0.905 0.173 0.000 14 0.267 -0.156 -0.470 1.000 -0.326 0.301 0.375 0.106 0.074 15 -0.846 -0.661 0.728 -0.326 1.000 0.521 -0.760 0.059 -0.270 16 -0.538 -0.702 0.562 0.301 0.521 1.000 -0.631 0.329 0.039 17 0.800 0.563 -0.905 0.375 -0.760 -0.631 1.000 -0.295 -0.062 18 0.021 -0.071 0.173 0.106 0.059 0.329 -0.295 1.000 0.561 19 0.178 0.270 0.000 0.074 -0.270 0.039 -0.062 0.561 1.000 CLASS 3 - weight 17 normalized weight 0.079 relative strength 3.75e-07 ******* class cross entropy w.r.t. global class 8.48e+00 ******* Model file: /home/tove/p/autoclass-c/data/glass/glass-mnc.model - index = 0 Numbers: numb/t = model term number; numb/a = attribute number Model term types (mtt): (multi_normal_cn MNcn) REAL ATTRIBUTE (t = R) |Mean-jk - numb t mtt description I-jk Mean StDev Mean-*k|/ Mean StDev t a -jk -jk StDev-jk -*k -*k 00 14 R MNcn Log Al: Wt.% Aluminu 0.942 ( 5.97e-01 3.12e-01) 9.31e-01 ( 3.06e-01 3.70e-01) m oxide 00 19 R MNcn Log Fe: Wt.% Iron ox 0.942 (-5.06e+00 1.04e+00) 8.50e-01 (-4.18e+00 1.66e+00) ide 00 13 R MNcn Log Mg: Wt.% Magnesi 0.942 (-2.24e+00 3.19e+00) 6.71e-01 (-9.37e-02 2.60e+00) um oxide 00 17 R MNcn Log Ca: Wt.% Calcium 0.942 ( 2.29e+00 2.30e-01) 4.51e-01 ( 2.18e+00 1.45e-01) oxide 00 18 R MNcn Log Ba: Wt.% Barium 0.942 (-4.94e+00 1.57e+00) 3.22e-01 (-4.43e+00 1.93e+00) oxide 00 16 R MNcn Log K: Wt.% Potassi 0.942 (-2.05e+00 2.83e+00) 2.05e-01 (-1.47e+00 1.74e+00) um oxide 00 15 R MNcn Log Si: Wt.% Silicon 0.942 ( 4.28e+00 1.65e-02) 1.88e-01 ( 4.29e+00 1.07e-02) oxide 00 11 R MNcn Log RI: refractive i 0.942 (-6.56e-01 5.89e-03) 1.39e-01 (-6.57e-01 5.82e-03) ndex 00 12 R MNcn Log Na: Wt.% Sodium 0.942 ( 2.59e+00 8.22e-02) 7.93e-02 ( 2.59e+00 6.09e-02) oxide Correlation matrix (row & column indices are attribute numbers) 11 12 13 14 15 16 17 18 19 11 1.000 -0.220 0.148 -0.761 0.631 -0.331 0.834 -0.310 0.203 12 -0.220 1.000 0.081 0.087 -0.270 -0.562 -0.266 0.165 -0.078 13 0.148 0.081 1.000 -0.232 -0.005 -0.468 -0.148 0.262 0.093 14 -0.761 0.087 -0.232 1.000 -0.745 0.526 -0.790 0.545 -0.153 15 0.631 -0.270 -0.005 -0.745 1.000 -0.335 0.795 -0.558 0.208 16 -0.331 -0.562 -0.468 0.526 -0.335 1.000 -0.269 0.236 0.005 17 0.834 -0.266 -0.148 -0.790 0.795 -0.269 1.000 -0.582 0.154 18 -0.310 0.165 0.262 0.545 -0.558 0.236 -0.582 1.000 -0.059 19 0.203 -0.078 0.093 -0.153 0.208 0.005 0.154 -0.059 1.000 CLASS 4 - weight 14 normalized weight 0.066 relative strength 9.76e-08 ******* class cross entropy w.r.t. global class 1.03e+01 ******* Model file: /home/tove/p/autoclass-c/data/glass/glass-mnc.model - index = 0 Numbers: numb/t = model term number; numb/a = attribute number Model term types (mtt): (multi_normal_cn MNcn) REAL ATTRIBUTE (t = R) |Mean-jk - numb t mtt description I-jk Mean StDev Mean-*k|/ Mean StDev t a -jk -jk StDev-jk -*k -*k 00 17 R MNcn Log Ca: Wt.% Calcium 1.141 ( 2.51e+00 1.92e-01) 1.71e+00 ( 2.18e+00 1.45e-01) oxide 00 11 R MNcn Log RI: refractive i 1.141 (-6.44e-01 8.62e-03) 1.56e+00 (-6.57e-01 5.82e-03) ndex 00 13 R MNcn Log Mg: Wt.% Magnesi 1.141 (-3.53e+00 3.13e+00) 1.10e+00 (-9.37e-02 2.60e+00) um oxide 00 15 R MNcn Log Si: Wt.% Silicon 1.141 ( 4.27e+00 2.18e-02) 5.68e-01 ( 4.29e+00 1.07e-02) oxide 00 14 R MNcn Log Al: Wt.% Aluminu 1.141 ( 2.35e-02 5.01e-01) 5.64e-01 ( 3.06e-01 3.70e-01) m oxide 00 16 R MNcn Log K: Wt.% Potassi 1.141 (-2.64e+00 2.07e+00) 5.62e-01 (-1.47e+00 1.74e+00) um oxide 00 12 R MNcn Log Na: Wt.% Sodium 1.141 ( 2.57e+00 1.33e-01) 1.86e-01 ( 2.59e+00 6.09e-02) oxide 00 19 R MNcn Log Fe: Wt.% Iron ox 1.141 (-4.01e+00 1.97e+00) 8.27e-02 (-4.18e+00 1.66e+00) ide 00 18 R MNcn Log Ba: Wt.% Barium 1.141 (-4.43e+00 2.40e+00) 2.89e-03 (-4.43e+00 1.93e+00) oxide Correlation matrix (row & column indices are attribute numbers) 11 12 13 14 15 16 17 18 19 11 1.000 -0.569 -0.139 0.228 -0.556 0.275 0.640 0.212 0.442 12 -0.569 1.000 0.526 -0.091 -0.076 0.110 -0.757 -0.042 -0.366 13 -0.139 0.526 1.000 0.198 -0.485 0.325 -0.649 0.140 -0.429 14 0.228 -0.091 0.198 1.000 -0.612 0.808 -0.152 0.419 0.521 15 -0.556 -0.076 -0.485 -0.612 1.000 -0.743 0.110 -0.454 -0.307 16 0.275 0.110 0.325 0.808 -0.743 1.000 -0.211 0.371 0.512 17 0.640 -0.757 -0.649 -0.152 0.110 -0.211 1.000 -0.283 0.390 18 0.212 -0.042 0.140 0.419 -0.454 0.371 -0.283 1.000 0.181 19 0.442 -0.366 -0.429 0.521 -0.307 0.512 0.390 0.181 1.000 CLASS 5 - weight 12 normalized weight 0.057 relative strength 5.36e-05 ******* class cross entropy w.r.t. global class 8.97e+00 ******* Model file: /home/tove/p/autoclass-c/data/glass/glass-mnc.model - index = 0 Numbers: numb/t = model term number; numb/a = attribute number Model term types (mtt): (multi_normal_cn MNcn) REAL ATTRIBUTE (t = R) |Mean-jk - numb t mtt description I-jk Mean StDev Mean-*k|/ Mean StDev t a -jk -jk StDev-jk -*k -*k 00 13 R MNcn Log Mg: Wt.% Magnesi 0.996 ( 1.25e+00 1.46e-01) 9.20e+00 (-9.37e-02 2.60e+00) um oxide 00 14 R MNcn Log Al: Wt.% Aluminu 0.996 (-9.09e-02 4.87e-01) 8.14e-01 ( 3.06e-01 3.70e-01) m oxide 00 18 R MNcn Log Ba: Wt.% Barium 0.996 (-5.01e+00 1.07e+00) 5.42e-01 (-4.43e+00 1.93e+00) oxide 00 19 R MNcn Log Fe: Wt.% Iron ox 0.996 (-3.46e+00 2.14e+00) 3.35e-01 (-4.18e+00 1.66e+00) ide 00 16 R MNcn Log K: Wt.% Potassi 0.996 (-1.86e+00 1.28e+00) 2.99e-01 (-1.47e+00 1.74e+00) um oxide 00 11 R MNcn Log RI: refractive i 0.996 (-6.55e-01 7.68e-03) 2.21e-01 (-6.57e-01 5.82e-03) ndex 00 15 R MNcn Log Si: Wt.% Silicon 0.996 ( 4.29e+00 1.05e-02) 8.96e-02 ( 4.29e+00 1.07e-02) oxide 00 12 R MNcn Log Na: Wt.% Sodium 0.996 ( 2.60e+00 4.99e-02) 5.52e-02 ( 2.59e+00 6.09e-02) oxide 00 17 R MNcn Log Ca: Wt.% Calcium 0.996 ( 2.18e+00 5.09e-02) 2.57e-02 ( 2.18e+00 1.45e-01) oxide Correlation matrix (row & column indices are attribute numbers) 11 12 13 14 15 16 17 18 19 11 1.000 0.387 0.251 -0.197 -0.603 -0.304 0.571 0.007 -0.334 12 0.387 1.000 0.654 -0.669 -0.751 -0.689 0.144 0.101 -0.507 13 0.251 0.654 1.000 -0.191 -0.745 -0.521 -0.227 0.033 -0.538 14 -0.197 -0.669 -0.191 1.000 0.356 0.499 -0.423 -0.125 0.219 15 -0.603 -0.751 -0.745 0.356 1.000 0.549 -0.194 -0.026 0.433 16 -0.304 -0.689 -0.521 0.499 0.549 1.000 -0.326 0.016 0.209 17 0.571 0.144 -0.227 -0.423 -0.194 -0.326 1.000 0.011 0.150 18 0.007 0.101 0.033 -0.125 -0.026 0.016 0.011 1.000 0.303 19 -0.334 -0.507 -0.538 0.219 0.433 0.209 0.150 0.303 1.000 CLASS 6 - weight 8 normalized weight 0.037 relative strength 6.02e-04 ******* class cross entropy w.r.t. global class 9.60e+00 ******* Model file: /home/tove/p/autoclass-c/data/glass/glass-mnc.model - index = 0 Numbers: numb/t = model term number; numb/a = attribute number Model term types (mtt): (multi_normal_cn MNcn) REAL ATTRIBUTE (t = R) |Mean-jk - numb t mtt description I-jk Mean StDev Mean-*k|/ Mean StDev t a -jk -jk StDev-jk -*k -*k 00 13 R MNcn Log Mg: Wt.% Magnesi 1.066 ( 1.28e+00 8.12e-02) 1.70e+01 (-9.37e-02 2.60e+00) um oxide 00 15 R MNcn Log Si: Wt.% Silicon 1.066 ( 4.27e+00 4.90e-03) 2.32e+00 ( 4.29e+00 1.07e-02) oxide 00 12 R MNcn Log Na: Wt.% Sodium 1.066 ( 2.65e+00 3.03e-02) 1.74e+00 ( 2.59e+00 6.09e-02) oxide 00 19 R MNcn Log Fe: Wt.% Iron ox 1.066 (-4.89e+00 1.31e+00) 5.48e-01 (-4.18e+00 1.66e+00) ide 00 16 R MNcn Log K: Wt.% Potassi 1.066 (-2.51e+00 2.24e+00) 4.63e-01 (-1.47e+00 1.74e+00) um oxide 00 14 R MNcn Log Al: Wt.% Aluminu 1.066 ( 2.11e-01 3.84e-01) 2.47e-01 ( 3.06e-01 3.70e-01) m oxide 00 11 R MNcn Log RI: refractive i 1.066 (-6.56e-01 6.23e-03) 2.08e-01 (-6.57e-01 5.82e-03) ndex 00 17 R MNcn Log Ca: Wt.% Calcium 1.066 ( 2.15e+00 1.34e-01) 2.07e-01 ( 2.18e+00 1.45e-01) oxide 00 18 R MNcn Log Ba: Wt.% Barium 1.066 (-4.28e+00 2.22e+00) 6.73e-02 (-4.43e+00 1.93e+00) oxide Correlation matrix (row & column indices are attribute numbers) 11 12 13 14 15 16 17 18 19 11 1.000 -0.428 0.670 -0.843 -0.659 -0.691 0.877 0.090 -0.422 12 -0.428 1.000 0.063 0.234 -0.092 -0.076 -0.481 -0.387 0.708 13 0.670 0.063 1.000 -0.721 -0.652 -0.750 0.643 -0.092 0.092 14 -0.843 0.234 -0.721 1.000 0.705 0.809 -0.837 0.106 0.369 15 -0.659 -0.092 -0.652 0.705 1.000 0.712 -0.584 -0.120 0.054 16 -0.691 -0.076 -0.750 0.809 0.712 1.000 -0.657 0.268 0.131 17 0.877 -0.481 0.643 -0.837 -0.584 -0.657 1.000 0.050 -0.485 18 0.090 -0.387 -0.092 0.106 -0.120 0.268 0.050 1.000 -0.189 19 -0.422 0.708 0.092 0.369 0.054 0.131 -0.485 -0.189 1.000