نوع مقاله : علمی -پژوهشی
نویسندگان
دانشگاه صنعتی امیرکبیر
چکیده
کلیدواژهها
عنوان مقاله [English]
Data analysis helps us to understand how we should achieve the expected results, so as to achieve more accurate processes, it is necessary to choose an analyzing method that is the best one for our subject. In order to analyze surface samples of Parkam district based on four values of longitude and latitude of sampling points and grades of copper and Molybdenum, we use the three useful method of K-Nearest Neighbor (KNN), K-Means and Neural Networks. One of the important viewpoints in data mining to analyze and investigate high volume of data and samples with different characteristics is clustering viewpoint that itself include different methods and techniques. One of the most famous algorithms of clustering is KNN algorithm to estimate according to the training examples. In fact, it is a non-parametric method used for classification and regression in order to reach relationships among variables while K-Means algorithm tries to divide data in K clusters based on a distance criterion. Neural networks can be useful tools in pattern recognition while there is not much information available for interpretation. In present study, to simulate and estimate copper grade in porphyry copper system of Parkam located in Kerman province, different learning algorithms that are mentioned are compared and results are shown. In this paper, comparing the results of the three algorithms is our target to pave the way of researchers. The results show that KNN has more correlation in contrast of neural networks and K-Means so using KNN can be more effective to estimate copper grade. The advantage of using KNN method relative to other estimation methods in present study is providing a specified and accurate pattern for decision makers in industry to estimate grade.
کلیدواژهها [English]