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.

Geographic information positioning and partitioning method and device based on deep learning

Abstract: The invention provides a geographic information positioning partitioning method and device based on deep learning. According to the method, a Geology API is used as a mapping tool; the problem that split attributes cannot be selected originally is solved; on this basis, a foundation is provided, A pre-pruning method and a PEP pruning method are designed; a decision tree model of the accurate positioning partition is obtained by combining data and Geology API characteristics and improving based on a decision tree C4.5 algorithm; and according to a preset information gain rate solving algorithm,a split attribute is selected from the solved attribute values, a prediction result obtained through a preset deep learning model is further positioned and partitioned through a preset decision treemodel, accurate partition information is obtained, and the technical effect that the positioning partition precision is not high is achieved.

Regression tree for functional response : application in oceanology

Abstract: We consider here the problem of building a regression tree when the response variable is a curve. Following the work of Yu and Lambert (1999), we give a detailed analysis of an extension of regression trees to the functional case. The conditions required for the criterion to be used to construct the tree model in the functional context is discussed. This extension is applied to an oceanological problem where the objective is to predict the shape of salinity profiles using several explanatory environmental variables. Functional PCA is proposed in order to enlighten the interpretation of the tree-based model. Finally, bagging procedure allows to increase the accuracy of the functional regression tree and leads to a more stable model although the tree structure is lost.

Reducing Computational Complexity in Markov Decision Processes Using Abstract Actions

Abstract: In this paper we present a new approach for the solution of Markov decision processes based on the use of an abstraction technique over the action space, which results in a set of abstract actions. Markovian processes have successfully solved many probabilistic problems such as: process control, decision analysis and economy. But for problems with continuous or high dimensionality domains, high computational complexity arises because the search space grows exponentially with the number of variables. In order to reduce computational complexity, our approach avoids the use of the whole domain actions during value iteration, calculating instead over the abstract actions that really operate on each state, as a state function. Our experimental results on a robot path planning task show an important reduction of computational complexity.

Improving the Validity of Decision Trees as Explanations

Abstract: In classification and forecasting with tabular data, one often utilizes tree-based models. This can be competitive with deep neural networks on tabular data [cf. Grinsztajn et al., NeurIPS 2022, arXiv:2207.08815] and, under some conditions, explainable. The explainability depends on the depth of the tree and the accuracy in each leaf of the tree. Here, we train a low-depth tree with the objective of minimising the maximum misclassification error across each leaf node, and then ``suspend'' further tree-based models (e.g., trees of unlimited depth) from each leaf of the low-depth tree. The low-depth tree is easily explainable, while the overall statistical performance of the combined low-depth and suspended tree-based models improves upon decision trees of unlimited depth trained using classical methods (e.g., CART) and is comparable to state-of-the-art methods (e.g., well-tuned XGBoost).

Comparative Analysis of Decision Support Models for Faculty Performance Evaluation

Abstract: The faculty is an important asset to guarantee that an academic institution operates as expected. Performance evaluation is an important tool used to assess faculty efficiency in the workplace. The study focuses on the comparison of three different decision support models identifying the suited model to be used in the proposed faculty performance evaluation framework. A local community college provided the historical data and documents to the researchers. The researcher selected three suitable decision support models and used Weka for data analysis. The results of preliminary data analysis examined shows that the identified faculty performance evaluation criterion includes 75% of the National Budget Circular (NBC) criteria; 15% IPCR and 10% College Involvement and Participation (CIP). The comparative analysis criteria used in analyzing the decision tree would be utilized as the model in the knowledge-based decision support system. With regards to build time, both Random Tree and REP Tree resulted in 0 seconds while M5P has 0.23 seconds. Build time would affect the model efficiency in terms of resources needed for execution. REP Tree has the highest size of tree produced in the model. Since all the decision tree models have positive coefficients, it indicates that when the value of one variable increases, the value of the other variable also tends to increase. The results of comparing the decision support models in this study had identified potential suitability of a model in faculty performance evaluation. Furthermore, policies in the locale could be based on the logical decision trees presented in this study.