Instances: 985
Attributes: 3
MACD Histogram
MACD B/S Signal
Buy/Sell Decision
Test mode:evaluate on training data
=== Classifier model (full training set) ===
Sigmoid Node 0
Inputs Weights
Threshold -0.3116269715579534
Node 3 3.274473439437122
Node 4 -3.383164048576112
Sigmoid Node 1
Inputs Weights
Threshold 0.2409820685394542
Node 3 -4.420565987952903
Node 4 4.64828210539793
Sigmoid Node 2
Inputs Weights
Threshold -4.040957147476074
Node 3 0.7804868368378183
Node 4 -1.598294099172754
Sigmoid Node 3
Inputs Weights
Threshold -6.504255241529023
Attrib MACD Histogram -18.463633865266885
Attrib MACD B/S Signal -0.043694720894122255
Sigmoid Node 4
Inputs Weights
Threshold -8.806123515543858
Attrib MACD Histogram -17.522682964236022
Attrib MACD B/S Signal -0.046929639173773546
Class 1
Input
Node 0
Class -1
Input
Node 1
Class 0
Input
Node 2
Time taken to build model: 0.95 seconds
=== Evaluation on training set ===
=== Summary ===
Correctly Classified Instances 543 55.1269 %
Incorrectly Classified Instances 442 44.8731 %
Kappa statistic 0.0904
Mean absolute error 0.3296
Root mean squared error 0.4077
Relative absolute error 96.2216 %
Root relative squared error 98.5509 %
Total Number of Instances 985
=== Detailed Accuracy By Class ===
TP Rate FP Rate Precision Recall F-Measure ROC Area Class
0.13 0.052 0.693 0.13 0.219 0.539 1
0.96 0.859 0.537 0.96 0.689 0.552 -1
0 0 0 0 0 0.758 0
Weighted Avg. 0.551 0.463 0.604 0.551 0.455 0.549
=== Confusion Matrix ===
a b c <-- classified as
61 408 0 | a = 1
20 482 0 | b = -1
7 7 0 | c = 0
|
Above the layer structure of this result shown in figure 1. The input features are MACD Histogram and MACD B/S Signal, the output is Buy/Sell Decision which includes three classes 1,-1 and 0 standing for buy, sell and hold decision. The associated weight for hidden and output layer are all given.
Figure 1: ANN structure for MACD