Decision tree model

About: Decision tree model is a research topic. Over the lifetime, 2256 publications have been published within this topic receiving 38142 citations.

Equivalence classes of random Boolean trees

Abstract: An And/Or tree is a binary plane tree, with internal nodes labelled by logical connectives, and with leaves labelled by literals chosen in a fixed set of k variables and their negations. Pick up uniformly at random such a Boolean tree with n leaves, and consider the Boolean function it represents. Finally, let the size n of the trees tend to infinity. This process defines a random distribution on Boolean functions of k variables, named the Catalan tree distribution. It has long been studied in the literature, however quantitative results were obtained only by taking in a last step an infinite limit for k. In the present paper, we investigate the global model such that the number of variables k_n is a function of n. We describe the whole range of the probability distributions depending on the function k_n, as soon as it tends jointly with n to infinity. In this context, we exhibit a threshold M_n, equivalent to n / ln n, such that, when k_n becomes larger, then the probability distribution becomes stable. To study this model, we mainly use analytic combinatorics and we extend the Kozik's pattern theory, first developed for the Catalan tree model.

The Prediction of Delay Time Class Caused by CTCS-3 Onboard System Fault Based on Decision Tree

Abstract: The faults of train control system will lead to delay, which will affect the operational efficiency of the railway network. In this paper, the decision tree algorithm (CART) is used to predict the delay time level caused by CTCS-3 On-board System Fault, which takes the location of train failure, the fault component of CTCS-3 on-board system, the fault phenomenon of CTCS-3 on-board system as data features. In the natural language fault record, based on expert experience, extract the key features needed and grade the delay time. The selected features are put into the decision tree algorithm for classification and prediction, SMOTE algorithm is used to solve the problem of unbalanced number of categories, and grid search algorithm is used to adjust the model parameters. Finally, the output results of the algorithm are analyzed. The decision tree model yields a classification accuracy of 76% for the given data of fault feature and can be considered for delay time level prediction caused by CTCS-3 system fault. From the experimental results, the proposed method can be recommended for the prediction of the delay time level caused by CTCS-3 system fault.

Tree Based Classification of Firewall Logs to Dodge Intrusion

Abstract: With increase in internet users and traffic it has become really essential to analyse the incoming traffic from a network to prevent users from viruses and network attacks and provide a safe and hassle free internet experience. In this research we build a robust model to classify actions to be performed on log files by a firewall to prevent intrusion and other security related threats. Our proposed works use various machine learning models for this work. Our proposed work bested the existing work in the literature when compared over several evaluation metrics. We found the Decision Tree model to be the best performing with an error rate of 0.00245.