Performance Of Decision Tree Model Research Articles

Decision Tree based Data Modelling for First Detection of Thalassemia Major

Thalassemia is an inherited blood disease which lacks hemoglobin, the protein that is carrying oxygen to the body. The severe one is called Thalassemia Major which needs special care about blood transfusion. The use of rule-based method to create an inference as the first diagnosis of Thalassemia Major is not effective as rules have to be achieved from long interview with the medical personnel. This research aims to create a model based on decision tree for first detection of Thalassemia Major. The dataset is obtained by interview of Thalassemia symptoms and primary data of medical records from a hospital. Classical decision tree models used are ID3, C4.5 and CART. The models are evaluated by Train-Test Split consists of 70% training and 30% testing data and k-Fold Validation for checking model’s overfitting or underfitting. The output of this research is a final tree model from the best performance of decision tree models. The final result shows that C4.5 has the best performance with accuracy 100% and not overfitting or underfitting. Also, C4.5 performs feature selections to its tree modeling to simplify the inference. In brief, decision tree based modeling is effective to be used as first detection of Thalassemia Major by interview symptoms with generating automatic rules from its tree model.

Comparative Analysis of Various Machine Learning and Deep Learning Approaches for Car Resale Price Prediction in the Turkish Market

With escalating environmental concerns worldwide, the shift towards second-hand car markets has emerged as an eco-friendly alternative to reduce the carbon footprint associated with manufacturing new vehicles. However, the lack of accurate and efficient price prediction mechanisms may impede the growth and efficiency of these markets. This study, focusing on the Turkish second-hand car market, contributes towards addressing this gap by introducing a unique, comprehensive dataset gathered from various online markets across Turkey, thereby offering a broad spectrum of data pertaining to different vehicle types, specifications, and resale conditions. The study employs both classical machine learning methods, specifically decision trees, and deep learning models to predict used car prices. This comparative analysis aims to assess the potential of these methods in improving the predictability and transparency of resale price determination. Despite the superior performance of decision tree models, the study found that deep learning techniques achieved comparable results, indicating their potential for further optimization and enhancement. The accurate prediction of resale prices could streamline the operations of second-hand car markets, increasing their appeal to potential buyers and sellers. This could, in turn, contribute to environmental sustainability by encouraging the usage of pre-owned vehicles, which can substantially reduce the demand for new cars and thereby lower the environmental impact associated with their production.

DTEXNet: Artificial Intelligence-Based Combination Scheme for DDoS Attacks Detection

Distributed Denial of Service (DDoS) attacks became the most widely spread attack because of their ease of design and execution. Such attacks are challenging to detect and mitigate due to the diversity of DDoS attack modes and the variable size of attack traffic. This makes the research on DDoS attack detection extremely important. Machine learning techniques are used to detect various DDoS attacks with complex and dynamic patterns. However, such techniques require extensive pre-processing and feature engineering to the data to achieve acceptable results. On the other hand, neural networks can achieve acceptable results without the need for such prior preparations. This paper proposes a novel combination scheme between EfficientNet, Xception, and Decision Tree models called DTEXNet. DTEXNet combines two neural networks to benefit from their ability to extract features without the need to prior preparations, and a classical machine learning model that has high performance on similar problems. The solution proposed uses two convolutional neural networks (CNNs) to classify between 10 types of DDoS attacks and uses their prediction results to enhance the performance of Decision Tree model on the same classification task. The results of the experiments carried out show that the proposed solution can significantly improve the results of the Decision tree, EfficientNet, and Xception models if applied individually.

Rapid land use prediction via portable X-ray fluorescence (pXRF) data on the dried lakebed of Avlan Lake in Turkey

The rapid, non-destructive, and low-cost characterization of soil by using portable X-ray fluorescence (pXRF) has been increasingly attractive among soil scientists for soil pedogenesis research. Therefore, pXRF can aid to characterize by measuring elemental concentration, weathering indices, soil spectra for agricultural and lakebed areas. This study aimed to characterize the agricultural area (AA) and lakebed area (LA) through elemental concentration, weathering indices, and soil spectra based on pXRF spectral behavior and to evaluate the prediction and characterization performance of decision tree models for AA and LA. In total 108 soil samples (36 locations) were collected from three fixed depth intervals such as 0–20, 20–40, and 40–60 cm in both AA and LA in Elmalı, Antalya, Turkey. All soil samples measured using pXRF to obtain elemental concentrations and soil spectra and results of input variables were randomly separated into 70% for calibration and 30% for validation. Savitzky-Golay Filter was applied for spectral data preprocessing. The area under the ROC curve (AUC) and validation accuracy were used to compare the prediction performance of seven different models. The results showed that the AA had the highest Mg, Si, Ca, Ti, P, Mn, Zn, Sr, and Zr concentrations compared with LA. The prediction performance of AUC (0.79) and validation accuracy (0.77) for elemental concentrations were acceptable, but other models were less satisfactory compared to the best model. In particular, P, Sr, and Fe were robust indicators to separate AA and LA using the decision tree model. These results suggested that the possibility of using pXRF for soil characterization and high prediction performance of AA and LA. This study highlights that elemental concentrations, weathering indices, and soil spectra based on pXRF measurements can be used as a useful proximal sensor to predict, characterize, separate, and investigate an understanding of soil characterization in AA and LA.

Depression Prediction amongst Chinese Older Adults with Neurodegenerative Diseases: A Performance Comparison between Decision Tree Model and Logistic Regression Analysis

Abstract Data analyses using artificial intelligence (AI) have not gained popularity in social work as much as other disciplines. To demonstrate its use, this study focused on Chinese older adults with neurodegenerative diseases (NDs) to (i) develop a prediction model using decision tree model to identify factors associated with depression and (ii) compare the prediction performance of decision tree model with that of logistic regression analysis. Decision tree model processing involved four stages: data collection, data preparation, model development, and result evaluation. An algorithm named Classification and Regression Trees (CARTs) was utilised to grow the decision tree by Python 3.7.1. The performance evaluation was based on accuracy, sensitivity, specificity and Goodness index (G). Seven factors grew the decision tree, including Instrumental Activities of Daily Living (IADLs), Mini-Mental State Examination (MMSE), Health status, Activity of Daily Living (ADL), Gender, Self-rated health change and Age. When compared to logistic regression, the decision tree model had a much better performance in depression prediction. Researchers, practitioners and policymakers need to focus on ways to decrease the vulnerability of depression in Chinese older adults with NDs. Also, the decision tree model can be applied as a referral to other physical or mental diseases prediction and analysis.

Prediction accuracy of underground blast variables: decision tree and artificial neural network

Accurate prediction of blast induced ground vibration variables such as particle velocity and frequency are of interest for safe design of controlled blasting operations for mining, tunnelling or excavation projects. There are certain limitations in the widely used empirical and numerical approaches especially when number of variables is large. Various data driven approaches have been employed for producing correct estimates for such cases. Decision tree (DT), earlier successfully employed for solving variety of civil engineering problems, is employed for prediction of blast variables for the first time in this article. The performance of DT models was found to be equally good (for particle velocity variable) or better than (for frequency variables) ANN models developed in this study, and unequivocally superior to the SVM or RF models reported in literature. Additionally, the clarity in decision rule-based estimation foster easy comprehension and future implementation.

Machine Learning Approach for Bottom 40 Percent Households (B40) Poverty Classification

Malaysia citizens are categorised into three different income groups which are the Top 20 Percent (T20), Middle 40 Percent (M40), and Bottom 40 Percent (B40). One of the focus areas in the Eleventh Malaysia Plan (11MP) is to elevate the B40 household group towards the middle-income society. Based on recent studies by the World Bank, Malaysia is expected to enter the high-income economy status no later than the year 2024. Thus, it is essential to clarify the B40 population through a predictive classification as a prerequisite towards developing a comprehensive action plan by the government. This paper is aimed at identifying the best machine learning models using Naive Bayes, Decision Tree and k-Nearest Neighbors algorithm for classifying the B40 population. Several data pre-processing task such as data cleaning, feature engineering, normalisation, feature selection: Correlation Attribute, Information Gain Attribute and Symmetrical Uncertainty Attribute and sampling methods using SMOTE has been conducted to the raw dataset to ensure the quality of the training data. Each classifier is then optimized using different tuning parameter with 10-Fold Cross Validation for achieving the optimal values before the performance of the three classifiers are compared to each other. For the experiments, a dataset from National Poverty Data Bank called eKasih obtained from the Society Wellbeing Department, Implementation Coordination Unit of Prime Minister's Department (ICU JPM), consisting of 99,546 households from 3 different states: Johor, Terengganu and Pahang are used to train each of the machine learning model. The experimental results using 10-Fold Cross-Validation method demonstrates that the overall performance of Decision Tree model outperformed the other models and the significance test specified the result is statistically significance.

Prediction of Surface Currents Using High Frequency CODAR Data and Decision Tree at a Marine Renewable Energy Test Site

Abstract In this study, Decision Tree (DT) was employed to predict surface currents in a ¼-scaled marine renewable energy test site—Galway Bay. In training and testing models, wind speed, wind direction and tidal water elevation from a forecasting model, and observations of surface velocity components during previous hours were taken as input variables; surface velocity components were taken as the output variable. Appropriate value of Complexity Parameter (CP) in decision tree models was determined based on experiments producing the minimum Root-Mean-Square-Error (RMSE) values compared with the radar data. Statistics including RMSE, bias, correlation (R) and Scatter Index (SI) were computed between predictions and radar data to assess predictions. Results indicated that the DT model can produce satisfactory predictions of surface currents. Good performance of DT model indicated that it can be regarded as a promising approach for future applications.

Prediction of defect severity by mining software project reports

With ever increasing demands from the software organizations, the rate of the defects being introduced in the software cannot be ignored. This has now become a serious cause of concern and must be dealt with seriously. Defects which creep into the software come with varying severity levels ranging from mild to catastrophic. The severity associated with each defect is the most critical aspect of the defect. In this paper, we intend to predict the models which will be used to assign an appropriate severity level (high, medium, low and very low) to the defects present in the defect reports. We have considered the defect reports from the public domain PITS dataset (PITS A, PITS C, PITS D and PITS E) which are being popularly used by NASA’s engineers. Extraction of the relevant data from the defect reports is accomplished by using text mining techniques and thereafter model prediction is carried out by using one statistical method i.e. Multi-nominal Multivariate Logistic Regression (MMLR) and two machine learning methods viz. Multi-layer Perceptron (MLP) and Decision Tree (DT). The performance of the models has been evaluated using receiver operating characteristics analysis and it was observed that the performance of DT model is the best as compared to the performance of MMLR and MLP models.

Physico-chemical and biological characterization of anopheline mosquito larval habitats (Diptera: Culicidae): implications for malaria control

BackgroundA fundamental understanding of the spatial distribution and ecology of mosquito larvae is essential for effective vector control intervention strategies. In this study, data-driven decision tree models, generalized linear models and ordination analysis were used to identify the most important biotic and abiotic factors that affect the occurrence and abundance of mosquito larvae in Southwest Ethiopia.MethodsIn total, 220 samples were taken at 180 sampling locations during the years 2010 and 2012. Sampling sites were characterized based on physical, chemical and biological attributes. The predictive performance of decision tree models was evaluated based on correctly classified instances (CCI), Cohen’s kappa statistic (κ) and the determination coefficient (R2). A conditional analysis was performed on the regression tree models to test the relation between key environmental and biological parameters and the abundance of mosquito larvae.ResultsThe decision tree model developed for anopheline larvae showed a good model performance (CCI = 84 ± 2%, and κ = 0.66 ± 0.04), indicating that the genus has clear habitat requirements. Anopheline mosquito larvae showed a widespread distribution and especially occurred in small human-made aquatic habitats. Water temperature, canopy cover, emergent vegetation cover, and presence of predators and competitors were found to be the main variables determining the abundance and distribution of anopheline larvae. In contrast, anopheline mosquito larvae were found to be less prominently present in permanent larval habitats. This could be attributed to the high abundance and diversity of natural predators and competitors suppressing the mosquito population densities.ConclusionsThe findings of this study suggest that targeting smaller human-made aquatic habitats could result in effective larval control of anopheline mosquitoes in the study area. Controlling the occurrence of mosquito larvae via drainage of permanent wetlands may not be a good management strategy as it negatively affects the occurrence and abundance of mosquito predators and competitors and promotes an increase in anopheline population densities.