Forecasting The Price Of Stock With Repeated Editing Nearest-Neighbor Rules

Nearest-Neighbor rules is one of most important nonparametric pattern recognition classifiers up to now, the theory of it has been widely studied by different researchers and many practitioners. The nearest neighbor (NN) classification rules consist of several methods such as 1-NN and k-NN, which according to certain dissimilarity measurements. Experimental results with real data from various domains of stock price forecasting in the securities business, traditional NN not only require excessive times but also lead to enormous computational burden, and the risk is too tremendous with wrong decision-making. Editing NN algorithms are very sensitive to the total number of prototypes considered. This paper investigates the possibility of modifying optimal editing to cope with a broader range of practical situations. Most introduced editing algorithms are presented in a unified form and their different properties are intuitively analyzed. Optimal properties of the editing algorithms have been considered and their behavior under the small sample size assumption has been studied and illustrated with real experiments. The result clearly indicate that improved editing techniques are required for the cases in which only small samples are available which, in practice, are unfortunately too often the case. This generally gives improved and more robust results. From a practical standpoint, the computational burden that improved estimates imply is worth the benefits in performance, and behavior of the algorithms directly depends on the number of iterations which in turn depends on the complexity of the problem. These ideas could be combined using alternative, specially adapted, error estimates to better tune the trade-off between the error-estimation reliability versus statistical-independence. Experimental results with real data show that, NN rules with repeated editing and incapable-decision algorithms are very useful to the numerous investors, when they envisage an opportunity of negotiable securities.