The Frequency Distribution Curve Fitting And Its Biomedical Applications

Fit of frequency distribution and its application in biomedicine This study is concerned with how to fit the frequency distribution of data with distributions in math and how to use distribution in biomedicine.Nowadays the researches of frequency distribution of data are almost limited to normal distribution, while other distributions are seldom studied. Many frequency distributions of data are not normal. This kinds of frequency distribution of data widely exist. Unimodal frequency distribution of data often occurs. In view of that we fit the frequency distribution of data with distributions in math. Distributions of continuous random variables are unimodal, such as weibull distribution, gamma distribution, beta distribution, cauchy distribution, logistic distribution, laplace distribution. Only in cauchy distribution are expected value and variance non-exist. By sas programming with SQL and MACRO we solve the fit of frequency distributions of data with these distributions. Meanwhile good-of-fit test is actualized so that the bound of parameters of distribution function is dwindling until the desired accuracy is arrived. The process is executed by the circle of macro. According to the result of goodness-of-fit test the best distribution of function is determinedWidely used in biomedicine is the research of fit of distribution of frequency of data . In the paper only the range of normal value and the description of frequency distribution are discussed. In theory the widely-used percentile method to determine the range of normal value is imperfected. Curve fitting method show some advantage in determining the range of normal value, but is difficult to calculate. By fit of distribution of frequency of data the distribution of functions in theory are found so that the curve fitting method is easily used by the distribution function. The description of frequency distribution of data is a must. We can describe the central tendency with the arithmetic mean and the dispersion with the standard deviation. But we cannot do it if frequency distribution of data is not normal. Generally we describe the central tendency of the skewness distribution with median and the dispersion with interquartile distance. If the distribution function is known, expected value and variance may be calculated. It stands to reason to describe the frequency distribution...