Abstarct: In the past to create a map class packing different effects on the surface of a field of study that is very time consuming and expensive was to be used. With the advent of satellites and satellite images a it by satellite images and their processing is done . But the problem is that the precision of the extracted result for class classification is dependent on the performance of the image processing method. Common methods Enterprises ranking in the past have been able to accomplish this in an appropriate manner. With the advent of new and advanced statistical and machine learning methods, is hoped that these methods can classified satellite images a better class operations conventional methods do. In this study, the class of land use classification method using advanced statistical and recruiting machines and their performance compared with conventional methods, are. Thus, the radiation Abrtyfy be used during and Collected within a specific geographic are collected. The data included 176 explanatory variable (band satellite) and a response variable that is 13 categories of vegetation in the study area is characterized by geographic coordinates to.Values of the explanatory variable has been removed by the AVIRIS sensor. to extract the data from the satellite imagery LAND SAT was used. The study area includes KFC (Kennedy Flight center) is located in the state of Florida in America. Advanced statistical methods studied in this thesis, including support vector machines and random forests are .Support Vector Machine has 4 different kernel function method is that each of them has its own parameters that. Choice of kernel functions and optimize its parameter values using the grid search done, it has a high computational cost. Another method studied in this thesis is a using random forests. Several sets of random forest method the decision tree for the model parameters and variable number of trees required. Further optimization of these two parameters to do the costly and time can be. These two methods are listed in the Advanced Materials and Methods conventional statistical quadratic discriminant analysis, linear discriminant analysis in the engineering sciences to the maximum likelihood method, the distance be Mahalonobis and called. The final model compared to conventional statistical methods, advanced statistical is obtained. The comparison of the four models are Gaussian kernel Support Vector Machine is. The model parameters and overall accuracy 0/958, gamma = 0/001 and cost = 1000 and also has a high sensitivity for all class . In other words, the classification class of for all 13 class will be doing well.