Provide a New Category of Data mining toEnhance the Accuracy and Speed
Provide a New Category of Data mining toEnhance the Accuracy and Speed
zeinab Ahmadmehrabi1
IOI Code
:
XYCZ-AAECC
Publication :
International Conference on Engineering and Applied Science(iceasconf.com)
Abstract :
In this paper, classification algorithms on different data in terms of speed and accuracy review. To increase the accuracy of prediction models, group training has been introduced as a promising approach. This method is based on the combined results of the various forecast models are formed. In this thesis, two models have been proposed in a model of training models using a dataset that contains a subset of the features are used wo9ith data models and forest genetic algorithm and the second model of neural network training combined with meta-heuristic algorithms and model I have used for the training data set. In fact, in this model are two distinct processes. A process for the production of different collections and other process meta-heuristic algorithms used to train the neural network is responsible .In models of forest genetic algorithms and algorithms known as meta-heuristic algorithms have been used to produce educational settings. Parameters affecting the data classification and evaluation criteria models in both sensitivity (TPR), transparency (TNR), accuracy and precision. But in the experiments and tests due to the large volume of data only to three important measure of precision, accuracy and sensitivity are discussed.
The genetic algorithm parameters that three parameters considered to be cut after the test in the range [0.7.1], mutation is considered negligible amount of less than 0.15. Sowing rate effect on local forests algorithm [2, 4] and sowing across the range of [2-1] has been set. For each compound average of 40 runs has been done. Examined data from reference data UCI KDD and machine learning have been selected.
Keywords : data mining classification accuracy neural network genetics algorithm forest algorithm