Application Of Data Mining Technology In The Research Of Harmaceutical Experiment

The purpose of data mining is to discover knowledge hidden, mysterious and in which people are interested from mass databases.For instance,we frequently do some relevant measurement experiments in practice,in order to investigate the relationship between an effect and corresponding factors, and obtain some discrete data.Therefore,the relation analysis between discrete data and measurement factors is quite an important task.By pharmaceutical experimental tests,we can obtain a large,mount of discrete data.So these experimental data take place in combination of various factors(such as drug concentration,the temperature required for drug effect and the soaking time of drug effect,and so on).However,it is difficult for us to discover the relationship of these data only by a priori knowledge and experience,because they include some implicit associated rules,This paper attempts to apply data mining techniques to such data so as to find some hidden but useful information as scientific and theoretical basis for pharmaceutical research and development.Obviously,we mainly use multi-dimensional data analysis or OLAP method in data mining.A variety of data operation techniques,such as slice,cut,rotation and the volume of drilling can be adopted in OLAP multi-dimensional structure. But these techniques would cause some one-sidedness for data. So this paper tells us how to deal with this problem.First, it is essential to establish relevant function models by regression analysis on every data subset after handling slices. Then,slice factors can be regarded as a unit to group all functional models,and variable factors in each set of model will be digged comprehensively with partial differential coefficients,which can eliminate the one-sidednesses Slicing Processing brings.This paper proposes a high dimensional data mining model due to practical demand and the property of the processed objects.By means of discrete data pre-processor,which deals with the pharmacy effects data by slice technique in high dimensional data analysis,and multi-factor model deduction, we could obtain the effective relevant function models on data subset of different data on slice factor.Then,by partial differential effect,we analyze the model group in each slice factor on every variables.The paper indicates the global changing regulation of various influential factors yielded by the pharmacy effects process.So we could mine the data hidden in the effective information in partial differential effect analysis.In the same time we could eliminate one-sidedness in the slice process.According to the results of excavations and the combination with practical constraint conditions,we get the practical optimal reference solution,which coincides with the data of experimental effects greatly in pharmaceutical research and development.Therefore,the reference solution earns recognition by staff of the pharmaceutical research and development.