242
X-Machines for Agent-Based Modeling: FLAME Perspectives
Test mode:evaluate on training data
=== Model and evaluation on training set ===
EM
==
Number of clusters selected by cross validation: 5
Cluster
Attribute
0
1
2
3
4
(0.02)
(0.27)
(0.29)
(0.05)
(0.37)
=============================================================================
itno
mean
7.9139 243.8688 106.8236
18.0812 406.5556
std. dev.
3.2163
41.6689
45.2054
9.313
54.2469
cell_number
mean
741.0967 2344.6184 2337.2981 2332.0886 2342.4794
std. dev.
571.6462 110.9634 139.8895 193.7086
71.0952
normal
mean
185.0694
0
0 475.5265
0
std. dev.
484.6112 239.2861
0.0001 937.0862 239.2861
attacked by drug a
mean
0.7637
0
0
4.4818
0
std. dev.
1.9997
2.5182
2.5182
10.4813
2.5182
attacked by drug b
mean
239.6292
0.0382
13.5826 415.5753
19.509
std. dev.
197.3125
1.3171
32.9128 269.2916
12.4946
nohope-resistant to both
mean
0
2.1379
0
0 2322.9703
std. dev.
1126.6877
68.9101
0.1567 1126.6877
71.0925
Time taken to build model (full training data) : 16.86 seconds
=== Model and evaluation on training set ===
Clustered Instances
0
8 ( 2%)
2
285 ( 57%)
3
20 ( 4%)
4
188 ( 38%)
Log likelihood: -20.56071
With differing values of standard deviations and likelihoods it is still difficult to spot wrong model conditions. However the graphs in Figure 8.3 shows
the conflicting picture with a spike in mutated cells suddenly with time. The
graph, in this case, is better at showing that the model has some wrong conditions leading to erratic behavior.