Gene Microarray Data Prediction Algorithm Research

Applications of Gene microarray data in the current gene prediction of diseases have a significant effect,it can help improve health care efficiency and improve health outcomes.However,with the development of the informationize process of medical area,medical industry faced challenges of massive data and unstructured data.A lot of data so numerous and diverse difficult to establish prediction model to predict the performance is not good.This article have analyzed and forecasted for Gene microarray data,for the following main studies:(1)This paper describes and discusses the important properties of microarray dataset and the methods of conducting field screening.Using these methods to extract important attributes of Gene microarray data inside and limit the choice of learning machine algorithm to filter out the sample data prediction model to identify the data mapping rules between labels and features to take advantage of the prediction model training was subjected to disease prediction.Extreme Learning Machine(ELM)is a very fast data mining algorithm using least square principle to establish the relationship between the expression between predicted and actual values,and then using the generalized inverse principle for solving the minimum norm satisfy the relationship of Least Squares Solutions of the forecast model is an important factor,to establish a final predictive model.Its generalization and solving problem speed than traditional neural network algorithm,decision trees,and Support Vector Machine(SVM)and other prediction algorithm.(2)The reason this paper is to optimize the forecasting model,during data sampling,data dimensions mapped into a high dimensional space,a portion of the sample data sample data set was mistakenly included in the other classifications,so unstructured microarray data the presence of different proportions of sample outliers,so extreme Learning Machine(ELM)generalization and prediction accuracy of the algorithm are affected.This article causes ELM(ELM)poor performance of the algorithm to make the analysis concluded that during the time the prediction model,the training algorithm to each sample were treated the same,there must be abnormal sample data impact forecast model the performance,in order to reduce the influence of erroneous sample data prediction model performance,the article proposes the input sample data are weighted solutions,optimization of the two algorithms,which are weighted extreme Learning Machine(WELM)and double-weighted the extreme Learning Machine(BWELM).(3)In this paper,University of California at Irvine(UCI)4 group of diseases microarray datasets machine learning library are provided on the traditional limits of learning machine with single,double weighted ultimate learning machine algorithm analysis to predict the experiments to an algorithm to predict the performance of each in different outliers(outliers increased from 0 to 0.3)is.The experimental results indicate that in the presence of outliers,after weighted prediction accuracy over the limit higher learning machine algorithm.