Information Technology Reference
In-Depth Information
Accuracy of Naive Bayes
Accuracy of Maximum Entropy
Precision of Naive Bayes
Precision of Maximum Entropy
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
100
200
300
400
500
600
700
800
Number of Training Data
Fig. 2.11
Accuracy and precision of classification methods through number of training data
The accuracy of the Naïve Bayes pursues nearly a straight line from 850 to 600
training data. It reaches its maximum level at 520 training data, and then follows a
slow decrease.
The precision level of the Naïve Bayes increases from 850 to 520 training data,
and reaches its optimum level at 520. Afterward, it demonstrates a decreasing trend.
2.4.2 Performance of Maximum Entropy by Iteration Number
In this section, how the MaximumEntropy classification method of NLTK is affected
by the change in the number of iterations is demonstrated. The number of iterations
is set as 10 at first, and then increased by 5 until 75. The training dataset and the
testing dataset are not changed, the only altered element is the iteration number of the
Maximum Entropy method. The Fig.
2.12
shows the accuracy and precision values
of the Maximum Entropy method.
As one can spot from the figure, the accuracy value of the Maximum Entropy
demonstrates an increasing pattern until the point where the iteration number reaches
25, and then it goes through a slight decrease; however, it increases until the point
where the iteration number indicates 45 where the accuracy value reaches its peak
level. After the peak, the value pursues a slowly decreasing path. After the iteration
number shows 50, the accuracy value of the Maximum Entropy does not change
almost at all regardless of the increase in the number of iterations.
The precision value of theMaximumEntropy demonstrates a steady increase until
the point where the number of iteration reaches 65 although its pace decreases for
some time. It shows discontinuation for a short period of time between the iteration
Search WWH ::
Custom Search