Study On Data Mining Technology By Genetic Algorithm

Data mining technology has been extensively used until now, but study on data mining algorithms is still on the up. Among so many data mining algorithms, forecasting and clustering algorithms are very common in use, and a lot of scholars have been doing research about them. Although new algorithms are put forward successively, most of these algorithms improve old algorithms from one or more aspect, none of them can declare it has improved all disadvantages in existing algorithms. This paper is also not a perfect resolution; the author gives new solutions for forecasting and clustering algorithms.In forecasting of data mining algorithms, regression analysis method is the most common way to do forecast research, and it is also very simple and easy in use. But this kind of analysis has some shortages in forecasting application; the result is often not precise enough, sometimes obvious departure even occurs. Genetic algorithm performs well in searching for global optimized peaks in resolving problems,so in this paper genetic algorithm is used to improve the performance of regression analysis.Clustering is another technology used frequently in data mining. K-means algorithm is the most common one among those algorithms. But in the application of this algorithm, it needs user to input the value of k aimed to cluster. However, users sometimes have no idea about this parameter, besides, different value of k results in different cluster although the algorithm is the same. Another disadvantage in this algorithm is that this algorithm needs to produce a random initial cluster center, but different initial cluster center has different result. Sometime if the initial cluster center has been chosen improperly, the result may be the local optimized result. So in this paper, the author gives a new method to improve this algorithm using genetic algorithm.At the end, this paper gives respectively examples for these two algorithms to show that new methods perform better than old ones in deed.