Decision Tree Structure Research Articles

DT-SNE: t-SNE discrete visualizations as decision tree structures

Visualizations are powerful tools that are commonly used by data scientists to get more insights about their high dimensional data. One can for example cite t-SNE, which is probably one of the most famous and widely-used visualization techniques. However, t-SNE is a nonlinear and non-parametric technique that makes it suffer from a lack of interpretability. In this paper, we present a new technique inspired by t-SNE’s objective function that combines its ability to build nice visualizations with the interpretability of decision trees. This new visualization technique, called DT-SNE, can be seen as a discrete visualization technique where groups of instances are provided, as well as a ranking between them. The decision rules of the decision tree provide clear insights to interpret these different groups.

A Decision Support Tool for Social Engagement, Alternative Financing and Risk Mitigation of Geothermal Energy Projects

This paper presents a decision support tool for promoters/investors of geothermal energy projects, based on a decision tree (DT) structure. The DT aims to assist stakeholders to select public engagement strategies, alternative financing solutions and risk mitigation measures (or options) for geothermal energy projects. Public engagement is necessary for the successful development and operation of geothermal projects. Available studies (including toolkits and protocols) commonly list a set of practices for social engagement without providing information on the factors which render certain options more suitable than others. The presented tool offers a transparent framework to how relevant decisions could be managed by providing a sequence of questions that focus on social, environmental, resource risk, and financial influencing factors and to realise community engagement into geothermal projects. This work is part of the Horizon 2020 CROWDTHERMAL project, which aims at empowering the public to directly participate in the development of geothermal projects through social engagement tools and alternative financing schemes, like crowdfunding.

Analysis of Teachers’ Cognitive Ability and Teaching Motivation on the Academic Achievement of Students in Higher Education via Employment Data Guidance

Teachers and students are the two basic elements in educational activities. Students are educated but are not exactly passive recipients of education. With subjective initiative, all educational impacts must be through the initiative of students to achieve the desired effect. Therefore, all activities of education must start from mobilizing students’ initiative and motivation so that they have sufficient motivation to learn actively and well. The effective analysis of employment data, at the statistical level of data analysis, is a favorable basis to support the influence of teachers on students. However, most of the previous methods are C4.5 algorithms, decision tree generation algorithms based on rough sets, etc., which are commonly used for employment data analysis. None of them can sufficiently deal with the problem of different decision accuracy requirements and noise adaptability. In this paper, we analyze the employment data of a university in 2012 as an example and compare the analysis results with those of the C4.5 algorithm and decision tree generation algorithm based on a rough set. The results show that the decision tree algorithm based on the multiscale rough set model generates a simple decision tree structure. In addition, our methods do not have indistinguishable datasets and are fast in terms of computing. This study provides an effective guide to the relevance of teachers’ cognitive abilities and teaching motivations for students’ employment.

A decision tree-based method for detecting middle school students' behaviour characteristics in online English learning

In order to solve the problem of low accuracy and long detection time caused by poor feature extraction effect of online English learning students' behavioural characteristics detection, this paper proposes a method of online English learning students' behaviour characteristics detection based on decision tree. Firstly, the concept and structure of decision tree are analysed, and the classification steps are designed. Secondly, weighted principal component analysis was used to extract the behaviour characteristics of students. Then, the characteristic data is standardised. Finally, the C4.5 decision tree algorithm is used to construct a student behaviour feature detection model to detect students' behaviour characteristics in online English learning. The experimental results show that the feature detection rate of the proposed method is as high as 99.5%, the accuracy is 96.2%, and the detection time is 19.8 s. Therefore, the feature detection effect of the proposed method is good, the accuracy is high, and the detection time is effectively shortened.

Consistency preserving database watermarking algorithm for decision trees

Database watermarking technologies provide an effective solution to data security problems by embedding the watermark in the database to prove copyright or trace the source of data leakage. However, when the watermarked database is used for data mining model building, such as decision trees, it may cause a different mining result in comparison with the result from the original database caused by the distortion of watermark embedding. Traditional watermarking algorithms mainly consider the statistical distortion of data, such as the mean square error, but very few consider the effect of the watermark on database mining. Therefore, in this paper, a consistency preserving database watermarking algorithm is proposed for decision trees. First, label classification statistics and label state transfer methods are proposed to adjust the watermarked data so that the model structure of the watermarked decision tree is the same as that of the original decision tree. Then, the splitting values of the decision tree are adjusted according to the defined constraint equations. Finally, the adjusted database can obtain a decision tree consistent with the original decision tree. The experimental results demonstrated that the proposed algorithm does not corrupt the watermarks, and makes the watermarked decision tree consistent with the original decision tree with a small distortion.

Deep learning for chemometric analysis of plastic spectral data from infrared and Raman databases

Increasing plastic recycling rates is key to addressing plastic pollution. New technologies such as chemometric analysis of spectral data have shown great promises in improving the plastic sorting efficiency to boost recycling rates. In this work, a novel deep learning architecture, PolymerSpectraDecisionNet (PSDN) was developed, consisting of convolutional neural networks, residual networks and inception networks in a decision tree structure. To better represent the conditions in the plastic recycling industry, the models were built to identify the most widely recycled polymers – polyethylene, polypropylene and polyethylene terephthalate from open-sourced infrared and Raman spectral dataset containing over 20 different polymers. PSDN performed better than end-to-end neural networks, obtaining an accuracy of 0.949 and 0.967 with the Raman and infrared datasets respectively. The use of deep learning can also distinguish between weathered and unaged polymer samples, with accuracies of 0.954 for high density polyethylene and 0.906 for polyethylene terephthalate.

Predictive Risk Modeling for Recurrence of Child Maltreatment Using Cases from the National Child Maltreatment Data System in Korea: Exploratory Data Analysis Using Data Mining Algorithm.

The purpose of this study was to explore and identify patterns of risk predictors of maltreatment recurrence using predictive risk modeling (PRM). This study used the administrative dataset from the National Child Maltreatment Information System recorded by Korean CPS (Child Protective Service) workers. The information, including recurrent maltreatment, was collected in 2012; then, those reported cases werefollowed for 2years through 2014. The data included information about child, family, caregiver, maltreatment, and service characteristics and consisted of male (50.22%) and female (49.78%) children with an average age of 9years (n = 4319). We examined the association of risk factors with recurrence using conditional inference trees (CTREE): a tree-based data mining algorithm for classification that allows the exploration of the interconnection between hypothesized risk factors. Study findings showed that a history of prior CPS involvement was the first decision point in the decision tree structure of recurrence. The effect of other risk factors depended on the presence of prior CPS involvement. In the absence of prior CPS involvement, cases with (a) a single-parent status and (b) a caregiver's alcohol abuse living in other types of households (two-parent households, kinship care, andchildren without parents) were associated with recurrence. In the presence of prior CPS involvement, cases with out-of-home care or others (long- or short-term foster care and emergency placement) in the final decision of child placement (a) where in-home care in the initial decision of child placement within the presence of physical abuse and (b) where social isolation without physical abuse was related to recurrence. Cases with (a) a male caregiver and (b) a female caregiver with social isolation and without social isolation yet employed were at high risk for recurrence under the circumstance of in-home care in the final decision of child placement. This exploratory study found multiple connections among the factors in the prediction of recurrence. The CTREE helps unravel the complexity embedded in maltreatment recurrence by capturing its patterns. This information can deepen our knowledge of associations between risk factors in the prediction of recurrence and be used as a reference to inform child maltreatment policy and prevention.

Factors Controlling a Synthetic Aperture Radar (SAR) Derived Root-Zone Soil Moisture Product over The Seward Peninsula of Alaska

Root-zone soil moisture exerts a fundamental control on vegetation, energy balance, and the carbon cycle in Arctic ecosystems, but it is still not well understood in vast, remote, and understudied regions of discontinuous permafrost. The root-zone soil moisture product (30 m resolution) used in this analysis was retrieved from a time-series P-Band (420–440 MHz) synthetic aperture radar (SAR) backscatter observations (August 2017 & October 2017). While similar approaches have been taken to retrieve surface (0 cm to 5 cm) soil moisture from L-Band (1.2 GHz) SAR backscatter, this is one of the first known attempts at reaching the root-zone in permafrost regions. Here, we analyze secondary factors (excluding primary factors, such as precipitation) controlling summer (August) soil moisture at depths of 6 cm, 12 cm, and 20 cm over a 4500 km2 area on the Seward Peninsula of Alaska. Using a random forest model, we quantify the impact of topography, vegetation, and meteorological factors on soil moisture distributions. In developing the random forest model, we explore a variety of feature scales (30 m, 60 m, 90 m, 120 m, 180 m, and 240 m), tune hyperparameters (the structure of individual decision trees making up the ensemble including the number and depth of trees), and perform the final feature selection using cross-validated recursive feature elimination. Results suggest that root-zone soil moisture on the Seward Peninsula is primarily controlled by vegetation at 6 cm, but deeper in the soil column topography and meteorological factors, such as predominant winter wind direction and summer insolation, play a larger role. The random forest model accounts for 40% to 60% of the variation observed (R2 = 0.44 at 6 cm, R2 = 0.52 at 12 cm, R2 = 0.58 at 20 cm). These results indicate that vegetation is the dominant control on soil moisture shallow in the soil column, but the impact of vegetation does not extend to deeper layers retrieved from P-Band SAR backscatter.

Machine learning compensation of fiber nonlinear noise

This paper extends previous studies work on the application of Machine Learning (ML) to distortion compensation in optical communication systems with optical fibers nonlinear coefficients (γ) that exceed current industry standards e.g. \(\gamma > 1.4W^{ - 1} km^{ - 1}\). To quantify the improvement afforded by ML methods under different transmission conditions the procedures are applied to a model of a typical single-frequency optical communication system with, a 3200 km fiber length, double polarization, and a 16-QAM modulation format. The performance of both transmitters and receivers that incorporate Neural Networks (NNs) are then examined over a wide range of \(\gamma\) values. In all cases considered, the system Q-factor is improved although the degree of enhancement is dependent on the signal to noise ratio. The ML structures that were investigated include Siamese neural networks (SNN) implemented at the receiver end as well as two-stage architectures that employ NNs at the transmitter together with a classifier at the receiver side. Classifiers ranging from simple decision tree structures to boosting, forests, extra trees, and Multi-layer perceptron (MLP) were further examined and found to provide significant enhancement for \(\gamma >\) \(4W^{ - 1} km^{ - 1}\). The optimal performance for highly nonlinear systems was achieved for two-stage systems with random forest or extra tree classifiers. Finally, empirical equations for each ML technique were derived that relate the Q-factor enhancement and the value of \(\gamma ,\) and the number of triplet terms that are input into the neural network. These results could potentially be employed to relax manufacturing constraints and accordingly reduce system costs.

Fast mode decision for CU coding based on CNN for the VVC standard

Versatile Video Coding (VVC) is the latest generation of the video coding standard. In VVC, the advanced quadtree with a nested multitype tree (QTMT) partition structure provides more flexible coding unit (CU) partition sizes compared with the quadtree (QT) decision tree structure applied in the previous High Efficiency Video Coding (HEVC) standard. This flexibility, achieved by the new QTMT partitioning improvement, considerably improves the coding performance while increasing the coding computational complexity caused mainly by the rate distortion optimization processing. To overcome the complexity issue, a fast deep intra QTMT decision tree approach based on a convolution neural network (CNN) is adopted to determine the QTMT depth decision of each 128 × 128 Coding Tree Unit (CTU). The proposed algorithm predicts both the BT depths at 32 × 32 CUs and the QT depths at 64 × 64 using trained CNNs designed for each structure instead of processing the RDcost. Experimental results prove that the suggested deep QTMT approach achieves an important complexity reduction of up to 55.51% compared with the original reference software VTM3.0, with an average of about 35% encoding time reduction accompanied by an insignificant loss in encoding performance.

DeepTLF: robust deep neural networks for heterogeneous tabular data

Although deep neural networks (DNNs) constitute the state of the art in many tasks based on visual, audio, or text data, their performance on heterogeneous, tabular data is typically inferior to that of decision tree ensembles. To bridge the gap between the difficulty of DNNs to handle tabular data and leverage the flexibility of deep learning under input heterogeneity, we propose DeepTLF, a framework for deep tabular learning. The core idea of our method is to transform the heterogeneous input data into homogeneous data to boost the performance of DNNs considerably. For the transformation step, we develop a novel knowledge distillations approach, TreeDrivenEncoder, which exploits the structure of decision trees trained on the available heterogeneous data to map the original input vectors onto homogeneous vectors that a DNN can use to improve the predictive performance. Within the proposed framework, we also address the issue of the multimodal learning, since it is challenging to apply decision tree ensemble methods when other data modalities are present. Through extensive and challenging experiments on various real-world datasets, we demonstrate that the DeepTLF pipeline leads to higher predictive performance. On average, our framework shows 19.6% performance improvement in comparison to DNNs. The DeepTLF code is publicly available.

An Economic Evaluation of the Adjuvanted Quadrivalent Influenza Vaccine Compared with Standard-Dose Quadrivalent Influenza Vaccine in the Spanish Older Adult Population.

Standard-dose quadrivalent influenza vaccines (QIV) are designed to provide protection against all four influenza strains. Adjuvanted QIV (aQIV), indicated for individuals aged 65+ years, combines MF59® adjuvant (an oil-in-water emulsion of squalene oil) with a standard dose of antigen, and is designed to produce stronger and longer immune response, especially in the elderly where immunosenescence reduces vaccine effectiveness. This study evaluated the cost-effectiveness of aQIV vs. egg-based standard-dose QIV (QIVe) in the elderly population, from the payer and societal perspective in Spain. A dynamic transmission model, which accounts for herd protection, was used to predict the number of medically attended infections in Spain. A decision tree structure was used to forecast influenza-related costs and benefits. Influenza-related probabilities of outpatient visit, hospitalization, work absenteeism, mortality, and associated utilities and costs were extracted from Spanish and European published literature. Relative vaccine effectiveness (rVE) was sourced from two different meta-analyses: the first meta-analysis was informed by laboratory-confirmed influenza studies only, resulting in a rVE = 34.6% (CI95% 2–66%) in favor of aQIV; the second meta-analysis included real world evidence influenza-related medical encounters outcomes, resulting in a rVE = 13.9% (CI95% 4.2–23.5%) in benefit of aQIV. All costs were expressed in 2021 euros. Results indicate that replacing QIVe with aQIV in the Spanish elderly population would prevent on average 43,664 influenza complicated cases, 1111 hospitalizations, and 569 deaths (with a rVE = 34.6%) or 19,104 influenza complicated cases, 486 hospitalizations, and 252 deaths (with a rVE = 13.9%). When the rVE of aQIV vs. QIVe is 34.6%, the incremental cost per quality adjusted life years (QALY) gained was €2240 from the payer; from the societal perspective, aQIV was cost saving compared with QIVe. If the rVE was 13.9%, the incremental cost per QALY was €6694 and €3936 from the payer and societal perspective, respectively. Sensitivity analyses validated the robustness of these findings. Results indicate that replacing QIVe with aQIV in the Spanish elderly population is a cost-effective strategy for the Spanish healthcare system.

Automatic Modulation Classification: Decision Tree Based on Error Entropy and Global-Local Feature-Coupling Network Under Mixed Noise and Fading Channels

A decision tree structure is proposed to handle automatic modulation classification under mixed noise and fading channels. In the first layer of the tree, IQ signals are categorized into three groups according to the error entropy. In the second layer, IQ signals preprocessed by Cauchy-Score functions are fed as patterns into the neural network where global and local features are abstracted by Transformer and CNN parallelly and concatenated to make a final decision. Monte-Carlo experiments demonstrate the superiority of the proposed method, strategy and network under mixed noise and fading channels.

A novel SVM-based decision framework considering feature distribution for Power Transformer Fault Diagnosis

International Electrotechnical Commission (IEC) proposed the IEC three-ratio method based on Dissolved Gas Analysis (DGA), which is one of the most effective tools for Power Transformer Fault Diagnosis (PTFD). However, the PTFD accuracy is generally limited because the classification boundary could be too stiff to classify samples located near the boundary. The Support Vector Machine (SVM) was applied to PTFD to improve diagnosis accuracy, while traditional SVM multi-classification methods and parameter optimization algorithms are subject to poor training efficiency. As a result, the SVM-based PTFD model is difficult to update frequently with the accumulation of fault data. A new SVM-based PTFD decision framework is proposed in this paper which can significantly boost the training efficiency and ensure the accuracy. In the proposed framework, a multi-step feature extraction process consisting of characteristic gas concentration and its ratios is applied. Based on the feature distribution of various samples, a proper SVM multi-classification method is presented using a hierarchical decision tree structure. In addition, according to the principles of SVM and radial basis kernel function, a Support Vector feature-based parameter optimization algorithm (SVFB) is proposed. IEC TC 10 data and the historical data of online transformer monitoring provided by the State Grid Corporation of China are adopted as sample sets. The simulation results demonstrate that the proposed decision framework can reach high diagnosis accuracy while shortening the training time.

Forecasting and classification of new cases of COVID 19 before vaccination using decision trees and Gaussian mixture model

Regarding the pandemic taking place in the world from the spread of the Coronavirus pandemic and viral mutations, the need has arisen to analyze the epidemic data in terms of numbers of infected and deaths, different geographical regions, and the dynamics of the spread of the virus. In China, the total number of reported infections is 224,659 on June 11, 2022. In this paper, the Gaussian Mixture Model and the decision tree method were used to classify and predict new cases of the virus. Although we focus mainly on the Chinese case, the model is general and adapted to any context without loss of validity of the qualitative results. The Chi-Squared (χ2) Automatic Interaction Detection (CHAID) was applied in creating the decision tree structure, the data has been classified into five classes, according to the BIC criterion. The best mixture model is the E (Equal variance) with five components. The considered data sets of the world health organization (WHO) were used from January 5, 2020, to 12, November 2021. We provide numerical results based on the Chinese case.

Application of Ontology Matching Algorithm Based on Linguistic Features in English Pronunciation Quality Evaluation.

Traditional English classroom teaching is difficult to meet the oral learning needs of most learners. Thanks to the continuous advancement of speech processing technology, computer-assisted language learning systems are becoming more intelligent, not only pointing out learners' pronunciation errors but also assessing their overall pronunciation level. Method. This paper uses the method of tree kernel function to measure the similarity of two ontology trees. According to the features of nodes in ontology tree, methods to calculate the external features and internal features of nodes are proposed, respectively. External features are mainly obtained by calculating the hierarchical centrality, node density, and node coverage of nodes in the ontology tree; internal features are mainly obtained by measuring the richness of internal information. According to the similarity of ontology tree and the external features and internal features of nodes, the calculation formula of structural comprehensive similarity is improved, and the features of ontology itself can be fully considered in the calculation. According to the difference of the structure, the weights of the corresponding features during the calculation are adjusted autonomously, so that the calculation results are closer to reality. In spectral image preprocessing, endpoint detection utilizes the harmonic characteristics presented by narrowband spectrograms with high frequency resolution and eliminates useless nonspeech segments by detecting the presence of voiced segments. When building the neural network model, four convolutional layers, two fully connected layers, and one softmax output layer were conceived, and dropout was used to randomly suspend the work of some neurons to avoid overfitting. Results/Discussion. Through the data analysis of mean and variance and verified by one-way analysis of variance, it proves that the sentiment evaluation method in this paper is effective. The traditional multiple linear regression method is not suitable for the corpus and application scenarios of this paper. This paper proposes a decision tree structure, which is similar to the overall scoring process of raters, and uses the Interactive Dicremiser version 3 (ID3) algorithm to build a comprehensive evaluation decision tree for pitch, rhythm, intonation, speech rate, and emotion indicators. It is proved by experiments that the accurate consistency rate of the human-machine evaluation in this paper is 93%, the adjacent consistency rate is 96%, and the Pearson correlation coefficient value of the human-machine evaluation results is 0.89. The data results prove that the evaluation method in this paper is credible.

Research on Evaluation Method of Physical Education Teaching Quality in Colleges and Universities Based on Decision Tree Algorithm

Ensuring and improving quality in higher education is a never-ending theme in university construction and development. College teaching quality includes the quality of physical education instruction. Physical education’s unique focus necessitates the use of teaching quality evaluation tools. This paper develops a decision tree algorithm-based evaluation index system for college physical education teaching quality. The evaluation index and weight of the system for evaluating the quality of physical education teaching are determined using the expert consultation method and analytic hierarchy process. To put it another way, while discussing data mining theory in general, the paper focuses on decision tree algorithm flow and structure. To develop a decision tree model for the established PE teaching quality evaluation system, the modified C4.5 algorithm was applied to classroom teaching evaluation data from 125 teachers, and the model’s correctness and applicability were verified. When it is all said and done, it highlights key decision-making attributes that have an impact on how well physical education teachers are evaluated, and it offers some sound advice based on these findings.

Chatbot use cases in the Covid-19 public health response.

ABSTRACTObjectiveTo identify chatbot use cases deployed for public health response activities during the Covid-19 pandemic.Material and MethodsWe searched PubMed/MEDLINE, Web of Knowledge, and Google Scholar in October 2020 and performed a follow-up search in July 2021. We screened articles based on their abstracts and keywords in their text, reviewed potentially relevant articles, and screened their references to (a) assess whether the article met inclusion criteria and (b) identify additional articles. Chatbots, their use cases, and chatbot design characteristics were extracted from the articles and information from other sources and by accessing those chatbots that were publicly accessible.ResultsOur search returned 3334 articles, 61 articles met our inclusion criteria, and 61 chatbots deployed in 30 countries were identified. We categorized chatbots based on their public health response use case(s) and design. Six categories of public health response use cases emerged comprising 15 distinct use cases: risk assessment, information dissemination, surveillance, post-Covid eligibility screening, distributed coordination, and vaccine scheduler. Design-wise, chatbots were relatively simple, implemented using decision-tree structures and predetermined response options, and focused on a narrow set of simple tasks, presumably due to need for quick deployment.ConclusionChatbots’ scalability, wide accessibility, ease of use, and fast information dissemination provide complementary functionality that augments public health workers in public health response activities, addressing capacity constraints, social distancing requirements, and misinformation. Additional use cases, more sophisticated chatbot designs, and opportunities for synergies in chatbot development should be explored.

Random Kernel Forests

Random forests of axis-parallel decision trees still show competitive accuracy in various tasks; however, they have drawbacks that limit their applicability. Namely, they perform poorly for multidimensional sparse data that is frequently input in many text and image processing tasks. A straightforward solution is to create forests of decision trees with oblique splits; however, most training approaches have low performance or are hard to tune. Besides, those ensembles appear unstable and easily overfit, so they should be regularized to find the trade-off between complexity and generalization. This paper proposes an algorithm to train kernel decision trees and random forests. At the top level, this algorithm follows a common greedy procedure to train binary decision trees; however, it produces quasi-optimal linear and kernel splits at the decision stump level. At each stump, the algorithm finds a quasi-optimal distribution of classes to subtrees and trains this stump via optimization of an SVM-like loss function with a margin re-scaling approach, which helps simultaneously optimize the margin between subtree data and arbitrary impurity criteria. We also try to reveal uniform stability-based generalization bounds for those forests and apply them to select the regularization technique. The obtained bounds explicitly consider primary hyperparameters of forests, trees, and decision stumps. The bounds also show a relationship between outliers in decision tree structure, stability and generalization, and how forests smooth these outliers. We performed an experimental evaluation on several tasks, such as studying the reaction of social media users, image recognition, and bank scoring. The experimental results show that the proposed algorithm trains ensembles, which outperform other oblique or kernel forests in many commonly-recognized datasets. Namely, the proposed algorithm shows 99.1% accuracy on MNIST and 58.1% on CIFAR-10. It has been confirmed that the selected regularization technique helps reduce overfitting on several datasets. Therefore, the proposed algorithm may be considered a small step toward customized and regularized ensembles of oblique or kernel trees that keep reasonable training performance on large datasets.We believe that the proposed algorithmcan be utilized as a construction element of approaches to training context-aware models on complex problems with integer inputs and outputs, such as text mining and image recognition.

DroneSense: The Identification, Segmentation, and Orientation Detection of Drones via Neural Networks

The growing ubiquity of drones has raised concerns over the ability of traditional air-space monitoring technologies to accurately characterise such vehicles. Here, we present a CNN using a decision tree and ensemble structure to fully characterise drones in flight. Our system determines the drone type, orientation (in terms of pitch, roll, and yaw), and performs segmentation to classify different body parts (engines, body, and camera). We also provide a computer model for the rapid generation of large quantities of accurately labelled photo-realistic training data and demonstrate that this data is of sufficient fidelity to allow the system to accurately characterise real drones in flight. Our network will provide a valuable tool in the image processing chain where it may build upon existing drone detection technologies to provide complete drone characterisation over wide areas.