Classification by Nearest Neighbor and Multilayer Perceptron a New Approach Based on Fuzzy Similarity Quality Measure: A Case Study

In this paper the performance of k Nearest Neighbors and Multilayer Perceptron algorithm the is used in a classical task in the branch of the Civil Engineering: predict the level of service in the road. The use of fuzzy similarity quality measure method for calculating the weights of the features allows to performance of KNN and MLP in the case of mixed data (features with discrete or real domains). Experimental results show that this approach is better than other methods used to calculate the weight of the features. The results of the predictions of the level of service show the effectiveness of the method in the solution of problems of traffic engineering.