Project

General

Profile

Files » RFTBstatus_results_10_7_212TB_lipids_RF_raw_batch_1_train_only_15perc_filtered.txt

Random Forests results dataset 1 - Katie Lennard, 10/17/2018 09:28 AM

 
1
0 samples did not have response variable data, removing these...
2
Data set size:  26 samples with 16 and 10 samples per class
3
******************************************************************************
4
Building RF model on training set.....
5
Performing model tuning: 3 round(s) of 10 fold cross-validation
6
******************************************************************************
7
Confusion Matrix and Statistics
8

    
9
          Reference
10
Prediction active latent
11
    active     16      4
12
    latent      0      6
13
                                        
14
               Accuracy : 0.846         
15
                 95% CI : (0.651, 0.956)
16
    No Information Rate : 0.615         
17
    P-Value [Acc > NIR] : 0.0101        
18
                                        
19
                  Kappa : 0.649         
20
 Mcnemar's Test P-Value : 0.1336        
21
                                        
22
            Sensitivity : 1.000         
23
            Specificity : 0.600         
24
         Pos Pred Value : 0.800         
25
         Neg Pred Value : 1.000         
26
             Prevalence : 0.615         
27
         Detection Rate : 0.615         
28
   Detection Prevalence : 0.769         
29
      Balanced Accuracy : 0.800         
30
                                        
31
       'Positive' Class : active        
32
                                        
33
******************************************************************************
34
Cross-validated error rates associated with stepwise reduction of features:
35
   323    162     81     40     20     10      5      3      1 
36
0.1923 0.1923 0.2308 0.2692 0.3077 0.2692 0.2308 0.2692 0.4231 
37
******************************************************************************
38
THE TOP 10  MOST IMPORTANT FEATURES WERE:
39
                         predictors MeanDecreaseAccuracy
40
X864.947564511897 X864.947564511897                2.583
41
X504.306060791016 X504.306060791016                2.521
42
X580.308837890625 X580.308837890625                2.477
43
X716.28369140625   X716.28369140625                2.452
44
X917.193060876571 X917.193060876571                2.388
45
X1375.06449869989 X1375.06449869989                2.359
46
X468.984405517578 X468.984405517578                2.358
47
X482.975708007812 X482.975708007812                2.270
48
X940.994222415569 X940.994222415569                2.267
49
X686.960811605951 X686.960811605951                2.252
50
******************************************************************************
51
******************************************************************************
52
TRAINING SET classification summary if using the TOP 7 features only
53
******************************************************************************
54
Feature(s) selected: X864.947564511897
55
Feature(s) selected: X504.306060791016
56
Feature(s) selected: X580.308837890625
57
Feature(s) selected: X716.28369140625
58
Feature(s) selected: X917.193060876571
59
Feature(s) selected: X1375.06449869989
60
Feature(s) selected: X468.984405517578
61
Confusion Matrix and Statistics
62

    
63
          Reference
64
Prediction active latent
65
    active     16      5
66
    latent      0      5
67
                                        
68
               Accuracy : 0.808         
69
                 95% CI : (0.606, 0.934)
70
    No Information Rate : 0.615         
71
    P-Value [Acc > NIR] : 0.0308        
72
                                        
73
                  Kappa : 0.552         
74
 Mcnemar's Test P-Value : 0.0736        
75
                                        
76
            Sensitivity : 1.000         
77
            Specificity : 0.500         
78
         Pos Pred Value : 0.762         
79
         Neg Pred Value : 1.000         
80
             Prevalence : 0.615         
81
         Detection Rate : 0.615         
82
   Detection Prevalence : 0.808         
83
      Balanced Accuracy : 0.750         
84
                                        
85
       'Positive' Class : active        
86
                                        
(4-4/7)