CHAID Decision Tree Research Articles

Chi-Squared Automatic Interaction Detection Decision Tree Analysis of Social Determinants for Low Birth Weight in Virginia.

This study provides additional context to the literature regarding the social inequities that impact birth outcomes in Virginia using a decision tree analysis. Chi-squared automatic interaction detection data analysis (CHAID) was performed using data from the Virginia birth registry for the years 2015-2019. Birth weight was the outcome variable, while sociodemographic factors and maternity care deserts were the explanatory variables. The prevalence of low birth weight in Virginia was of 8.1%. The CHAID decision tree model demonstrated multilevel interaction among risk factors with three levels, with a total of 34 nodes. All the variables reached significance in the model, with race/ethnicity being the first major predictor variable, each category of race and ethnicity having different significant predictors, followed by prenatal care and maternal education in the next levels. These findings signify modifiable risk factors for low birth weight, in prioritizing efforts such as programs and policies. CHAID decision tree analysis provides an effective approach to detect target populations for further intervention as pathways derived from this decision tree shed light on the different predictors of high-risk population in each of the race/ethnicity demographic categories in Virginia.

Dropout and Graduation in Higher Education: CHAID Analysis

This study aims to investigate the socioeconomic variables and their order of importance that have a significant effect on the dropout and graduation of higher education students. Relational survey model was used in the study. In the study, the "Students Dropout and Academic Success Dataset," was utilized. The dataset, created by the Polytechnic Institute of Portalegre, consists of 4424 records. CHAID decision tree algorithm was used to analyze the data. With this method, the independent variables that demonstrate the maximum variation in the dependent variable have been identified hierarchically. It is found that, 49.93% of the students are “graduate”, 32.12% are “dropout”, and 17.948% are “enrolled”. Obtained findings show that the graduation rates of the students are not at the desired level. “Tuition fees up to date” was found as the best variable that explains the students’ school completion. 86.55% of students with not up to date tuition fees were found as dropout and 55.95% of students with up-to-date tuition fees were found as graduate. “Scholarship holder” was found as the variable that best explains the clusters formed by variable “tuition fees up to date”. 89.00% of the students that don’t have their tuition fees up to date and don’t hold a scholarship dropout the school, while 78.44% of students that have their tuition fees up to date and holding a scholarship are graduate. Building on the results obtained from the study, several suggestions were proposed for coping with dropout problem and further guiding research on dropout.

Diagnostic efficacy of data mining method based on multimodal ultrasound for papillary thyroid carcinoma.

The incidence of papillary thyroid caracinoma (PTC) is increasing year by year. Logistic regression model and Chi-squared automatic interaction (CHAID) decision tree based on multimodal ultrasound were established, and the diagnostic efficiency of the two models in PTC was compared. The findings, features and data of routine ultrasound, shear wave elastography (SWE) and contrast-enhanced ultrasonography (CEUS) were prospectively collected in 203 patients. Including: echogenicity, aspect ratio, maximum diameter of tumor, boundary, morphology, focal hyperecho, blood flow grading, maximum elasticity (Emax), minimum elastcity (Emin), mean elasticity (Emean), enhancement degree, enhanced characteristics, distribution of contrast agent, contrast medium arrival time. According to the pathological results, they were divided into PTC group and non-PTC group. CHAID decision tree model and binary Logistic regression model were established, receiver operator characteristic (ROC) curves of the two models were drawn, and diagnostic effectiveness was evaluated by comparing area under curve (AUC). Logistic regression showed that hypoechoic or very hypoechoic, aspect ratio ≥1, microcalcification and high SWE value were risk factors for PTC (OR 8.604, 2.154, 2.297, 1.067, respectively, P < 0.05). The CHAID decision tree showed echo, aspect ratio, Emax, contrast agent distribution and infusion time combined to diagnose PTC. ROC curve showed that the AUC of PTC predicted by Logistic regression model and CHAID decision tree model was 0.878 and 0.883, respectively, with no statistical significance (z=0.325, P=0.7456). Both Logistic regression model and CHAID decision tree model can play a good role in the diagnosis of PTC based on multi-modal ultrasound, but the diagnostic efficiency of both models is comparable. In conclusion, these two models provide new insights and ideas for PTC diagnosis.

Determining an Optimal Oxygen Saturation Target Range Based on Neonatal Maturity: Demonstration of a Decision Tree Analytic

The utility of decision tree machine learning in exploring the interactions among the SpO2 target range, neonatal maturity, and oxemic-risk is demonstrated. METHODS: This observational study used 3 years of paired age-SpO2-PaO2 data from a neonatal ICU. The CHAID decision tree method was used to explore the interaction of postmenstrual age (PMA) on the risk of extreme arterial oxygen levels at six different potential SpO2 target ranges (88–92%, 89–93%, 90–94%, 91–95%, 92–96% and 93–97%). Risk was calculated using a severity-weighted average of arterial oxygen outside the normal range for neonates (50–80 mmHg). RESULTS: In total, 7500 paired data points within the potential target range envelope were analyzed. The two lowest target ranges were associated with the highest risk, and the ranges of 91–95% and 92–96% were associated with the lowest risk. There were shifts in the risk associated with PMA. All the target ranges showed the lowest risk at ≥42 weeks PMA. The lowest risk for preterm infants was within a target range of 92–96% with a PMA of ≤34 weeks. CONCLUSIONS: This study demonstrates the utility of decision tree analytics. These results suggest that SpO2 target ranges that are different from typical range might reduce morbidity and mortality. Further research, including prospective randomized trials, is warranted.

Next-generation cyber attack prediction for IoT systems: leveraging multi-class SVM and optimized CHAID decision tree

Billions of gadgets are already online, making the IoT an essential aspect of daily life. However, the interconnected nature of IoT devices also leaves them open to cyber threats. The quantity and sophistication of cyber assaults aimed against Internet of Things (IoT) systems have skyrocketed in recent years. This paper proposes a next-generation cyber attack prediction framework for IoT systems. The framework uses the multi-class support vector machine (SVM) and the improved CHAID decision tree machine learning methods. IoT traffic is classified using a multi-class support vector machine to identify various types of attacks. The SVM model is then optimized with the help of the CHAID decision tree, which prioritizes the attributes most relevant to the categorization of attacks. The proposed framework was evaluated on a real-world dataset of IoT traffic. The findings demonstrate the framework's ability to categorize attacks accurately. The framework may determine which attributes are most crucial for attack categorization to enhance the SVM model's precision. The proposed technique focuses on network traffic characteristics that can be signs of cybersecurity threats on IoT networks and affected Network nodes. Selected feature vectors were also created utilizing the elements acquired on every IoT console. The evaluation results on the Multistep Cyber-Attack Dataset (MSCAD) show that the proposed CHAID decision tree can significantly predict the multi-stage cyber attack with 99.72% accuracy. Such accurate prediction is essential in managing cyber attacks in real-time communication. Because of its efficiency and scalability, the model may be used to forecast cyber attacks in real time, even in massive IoT installations. Because of its computing efficiency, it can make accurate predictions rapidly, allowing for prompt detection and action. By locating possible entry points for attacks and mitigating them, the framework helps strengthen the safety of IoT systems.

Hotel or rented flat: A profile of purchasers of accommodation services in Warsaw

The spreading of short-term flat rentals has brought about changes in the accommodation market, often seen as a threat to traditional accommodation providers. This is particularly true in large cities which have a considerable accommodation capacity and also a large stock of flats. The aim is to indicate to what extent short-term rentals are influencing the tourist accommodation market in Warsaw. The idea behind the study is the assumption that the differences revealed between those using hotels or such flats will provide an answer to the question of the influence of the latter on Warsaw»s tourist market. Such information should be useful in the marketing activities of interested parties and in the policies of the city authorities. Analysis of the data from a survey carried out in 2021 using the CHAID decision tree indicates that the choice of accommodation type was mainly determined by situational variables. The only statistically significant demographic predictor relates to a greater interest in flats among those aged up to 34 years old. Planned expenditure per person per overnight stay proved to be a statistically significant predictor only for non-residents of Poland, with the cut-off amount set higher than the median interval for this segment. Flats were more often chosen by people travelling in a larger party or alone and those planning to stay longer than four nights, thus looking for a different offer than that of traditional city hotels.

Influence of the pivot player on the effectiveness of back players in team-handball

In handball there is a consensus among the experts on the importance of the pivot player in the team's tactical effectiveness, not only because of his direct participation in the actions to finish off the attack, but also because of his influence on the actions of the rest of the players. In addition, playing in this specific position, most of the actions are carried out without the ball and take place inside the defensive system. These reasons make his influence on the outcome very difficult to analyze. In order to assess it, the present study has been carried out by collecting, through the observational instrument known as Influence of the pivot in the final situation, data from a sample of 1815 offensive sequences given in the medal matches of five top-level masculine tournaments. Different descriptive and correlational statistical analyzes were applied, among which the use of Chaid's decision tree technique stands out. It is concluded that the presence of the pivot in direct relation with the ball holder when the sequences are finished increases the probability that the back players will shoot, as well as increasing their effectiveness. Consequently, the influence of the pivot proves to be decisive for the game outcome.

Prediction of cognitive status and 5‐year survival for elderly with diabetes by using the Neurocognitive Frailty Index (NFI)

AbstractBackgroundType 2 diabetes mellitus has been shown to be associated to cognitive decline and an increased risk of mild cognitive impairment and dementia in elderly persons (Biessels et al., 2014). The aim of this research was to determine whether the Neurocognitive Frailty Index (NFI, Pakzad et al., 2017) could help to predict cognitive changes in elderly people with diabetes.MethodParticipants in this study were drawn from waves 1 (1991 to 1992) and 2 (1996 to 1997) of the Canadian Study of Health and Aging (CSHA) for prediction of 5‐year cognitive changes in those who received a consensus diagnosis of no cognitive impairment (NCI) or cognitive impairment but not dementia (CIND) in wave 1 (n = 997). The CIND category was contained to individuals whose level of cognitive impairment was measured to be greater than the NCI group but less than the dementia group. NFI was defined as a combined score of 41 physical and mental components (in 8 cognitive domains) as they were available in the dataset (NFI has 42 components but we withdraw diabetes). Cognitive score (measured by 3MS) at follow‐up, and mortality were outcome.ResultMean NFI and standard deviation was 9.7±6.22 in people without diabetes, 11.71±5.42 in people with diabetes and 9.36±5.30 in people with diabetes but are using treatment. Multiple linear regression indicated that NFI score is significantly associated to cognitive status (3MS at follow‐up, p&lt;0.05), however, diabetes did not show any significant association to predict this outcome. CHAID decision tree analysis stratified NFI scores into five categories, while proportional hazards analysis adjusted for age, gender and education indicated that NFI was more correlated to survival than age and had a dose‐response effect in relation to survival. Those with diabetes had a higher probability of death compared to those without (p&lt;0.05).ConclusionNFI was significantly associated with cognitive changes for the elderly in those with and without diabetes. Given that the NFI is proving to be a valuable tool to assess the risk of developing dementia, further investigation is needed to clarify which modifiable risk factors may help reduce the risk of cognitive decline.

Determinants of Subscription Renewal Behavior in Sport Spectatorship Services: A CHAID Decision Tree Modeling Approach

Abstract: Attempts were made to identify the prioritized determinants of season ticket holders’ (STHs) renewal behavior. Also, grounded in the lens of luxury fever, we explored how the determinants are differently weighed and processed across two types of STHs, including regular (R-STHs) vs. premium seat holders (P-STHs). Through a partnership with a Big Ten athletic department, the attendance data were obtained and analyzed. This study adapted a decision tree modeling approach to predict renewal/churning behavior by setting learning decision rules in the CHAID algorithm, a supervised learning algorithm family. Results of this study indicate that discounted tickets, affordability, and entertainment were important determinants for R-STHs. By contrast, tenure as a ticket holder, game outcomes, and business use were significantly weighed for P-STHs. Recent attendance and exclusive benefits were important branches for both decision trees. This research expands the understanding of the subscription-based services industry by applying a fresh data mining approach.

Applied machine learning analysis: Factors correlated with injection drug use and post-prison medication for opioid use disorder treatment engagement

Objectives This study aimed to classify the factors that were correlated with injection drug use (IDU) and with medications for opioid use disorder (MOUD) treatment engagement among individuals who were recently released from prison. Methods Data for this study were obtained from a Midwestern reentry program for incarcerated individuals with co-occurring opioid use and a mental health disorder between May 1, 2017, and April 30, 2020. CHAID decision tree modeling was utilized to classify IDU and MOUD treatment engagement. Results Those most likely to report IDU were individuals with a Hepatitis C diagnosis and a history of overdose, and those least likely to report IDU were not diagnosed with Hepatitis C, identified as a person of color, and never overdosed on opioids. The subgroup of that were most likely to report MOUD treatment engagement were individuals taking psychiatric medication and who had a history of IDU. The subgroup of participants least likely to report MOUD treatment engagement were individuals prescribed psychiatric medication, without had a history of IDU, and were not participating in substance-use treatment. Conclusion Our findings indicate that, to protect vulnerable populations and to flatten the overdose mortality curve, an increased focus is required within criminal/legal systems to facilitate linkages to care at reentry.

Predictive data mining models in the tests of propelling charges

In the article, in the introduction, the concept of predictive data mining models and was defined and the purpose of the article was specified. Then, the method of building predic-tive models was characterized and the elements of ammunition were indicated, the test results of which were prepared for the building of models, and the types of ammunition in which the propellant charge is present were indicated. The results of building four data mining models are presented. Predictive models for C&amp;RT, CHAID and exhaustive CHAID decision trees were designed and built. The fourth model analyzed was the SANN model, i.e. the model of neural networks. For each of the tree models, a schema of the designed tree, the rate of false predictions and the parameters of goodness of fit of the built models are shown. For the SANN model, the parameters of the selected neural network were addi-tionally characterized. An analysis of the built models was made and, based on the ob-tained results, the best designed predictive data mining model was indicated. At the end, the graphical form of the workspace predefined by the GC Advanced Comprehensive Classifiers project is shown.

Predictive Model of Functional Exercise Compliance of Patients with Breast Cancer Based on Decision Tree.

Regular functional exercise can help recover the functions of upper limb for patients with breast cancer. By finding the influencing factors of functional exercise compliance and constructing a predictive model, patients with a poor functional exercise compliance can be better identified. This study aims to find out the factors influencing the functional exercise compliance of patients with breast cancer and build a predictive model based on decision tree. Convenience sampling was used at two tertiary hospitals in Shantou from August 2020 to March 2021. Data of patients with breast cancer patient was obtained from questionnaires and based on demographics, Constant-Murley Score, Functional Exercise Compliance Scale for Postoperative Breast Cancer Patients, Champion Health Belief Model Scale, Social Support Rating Scale, Disease Perception Questionnaire and Family Care Index Questionnaire. Possible influencing factors of functional exercise compliance were analyzed using correlation analysis as well as univariate and binary logistic regression analysis through SPSS v25, and a CHAID decision tree was used to construct a predictive model on training, validation and test sets via SPSS Modeler v18 at a ratio of 6:2:2. Prediction accuracy, sensitivity, specificity and AUC were used to analyze the efficacy of the predictive model. A total of 227 valid samples were collected, of which 145 were assessed with a poor compliance (63.9%). According to a logistic regression analysis, perceived benefits, time after surgery and self-efficacy were influencing factors. The prediction accuracy, sensitivity, specificity and AUC of the predictive model, based on a CHAID decision tree analysis, were 70.73%, 57.1%, 77.8% and 0.81 respectively. The predictive model, based on a CHAID decision tree analysis, had a moderate predictive efficacy, which could be used as a clinical auxiliary tool for clinical nurses to predict patients' functional exercise compliance.

Financial Implications of Covid-19 Pandemic. An Empirical DuPont Analysis on Economic Value-Added Reported by European Developed Countries

The recent COVID-19 pandemic has raised various concerns related to firms’ financial vulnerabilities and resilience. The best way to achieve financial resilience is to identify drivers of sustainable economic development, mainly resumed to strategies designed aiming for shareholders’ value creation. The way firms’ resources are allocated indicate the premises for reaching the potential of value creation. This study provides a general picture of firms’ capacity to create value add. For this purpose, was performed a factorial analysis, considering a DuPont model. The analysis is performed on a sample of firms originating from well-developed European economies, considering a four-based DuPont factorial model analysis. For a better understanding of the contribution of each financial ratio of the DuPont financial analysis model, a CHAID decision tree was estimated. The results show generalized effects on the calculated economic valueadded indicator, no matter the country firms considered in the analysis are originated from, or the industry the operate in. However, in 2020, most of the firms have reported significantly better results in terms of economic value-added, so, an essential part of the discussion is related to the impact of savings made on the cost of the economic capital.

Klasifikasi Pemilih dalam Pemilu 2019 di Indonesia Menggunakan Regresi Logistik Multinomial dan Chi-Square Automatic Decision Tree (CHAID)

General Elections (Elections) in Indonesia are conducted once in every 5 years. This paper focused on the election of the president and his representatives. Voters in elections are regulated by law. There are many aspects of voter’s background that influence voters in making decisions during elections. In this paper, the focus is on voter background factors in geographical, demographic, and socio-economic aspects. The geographical aspect is the region where voters live; demographic aspects are gender, age, recent education, marital status, and religion; socioeconomic aspects are the household expenses, household income, personal expenses, type of work, type of residence, and status of residence. There are 3 categories of voter decisions in the election, namely candidate pair A, candidate pair B, and refuse to vote. The objectives of this paper are to find out what variables significantly influence voters in making decisions during elections, and to classify voters based on the significance of variables in geographical, demographic, and socio-economic aspects related to voter choice in elections. Multinomial logistic regression analysis is used to answer the first objective, while the CHAID (Chi-Square Automatic Decision Tree) decision tree is used to answer the second objective. Through multinomial logistic regression analysis, it can be seen that the variable type of region, age, recent education, religion, household expenditure, household income, house type and home status influence voter decisions in elections. Through the CHAID decision tree, the results obtained are 5 types of voters based on a decision tree of several independent variables that are significant to the dependent variable.

Validation of Individual Identification Through Decision Tree Packet Header Profiling

The drastic rise in the cybercrime rate associated with the surge of users' dependence on the Internet has elevated the concern of digital forensic examiners toward the footprints of perpetrators left in a virtual environment. However, suspect identification is a big challenge in network forensics due to the anonymous nature of data transmission across the network. This study utilises the decision tree classification approach to characterise users from their behavioural web navigation pattern using the meta-data of captured network packets (Destination IP, Protocol, Port Source, and Port Destination). A total of 95,795,379 network packet headers from 96 subjects were successfully collected. Their meta-data header packets were statistically profiled to generate digital fingerprints that try to link their action on the network to their identity accurately. Hence, CHAID decision tree modelling using Destination IP, Unique protocols, and a combination of the two, including Port source and Port destination, resulted in an accuracy of 4.07%, 6.34%, and 6.36%, respectively. However, the modelling could not create a reliable decision tree for the Port source and destination. The validation study on all the combined variables had a similar accuracy of 6.36%, indicating model created had reproducibility capability. Despite the outcome, the proposed method is not yet sufficiently strong for suspect identification. Further enhancement to improve its accuracy is required.

VALIDATION OF INDIVIDUAL IDENTIFICATION THROUGH DECISION TREE PACKET HEADER PROFILING

The drastic rise in the cybercrime rate associated with the surge of users' dependence on the Internet has elevated the concern of digital forensic examiners toward the footprints of perpetrators left in a virtual environment. However, suspect identification is a big challenge in network forensics due to the anonymous nature of data transmission across the network. This study utilises the decision tree classification approach to characterise users from their behavioural web navigation pattern using the meta-data of captured network packets (Destination IP, Protocol, Port Source, and Port Destination). A total of 95,795,379 network packet headers from 96 subjects were successfully collected. Their meta-data header packets were statistically profiled to generate digital fingerprints that try to link their action on the network to their identity accurately. Hence, CHAID decision tree modelling using Destination IP, Unique protocols, and a combination of the two, including Port source and Port destination, resulted in an accuracy of 4.07%, 6.34%, and 6.36%, respectively. However, the modelling could not create a reliable decision tree for the Port source and destination. The validation study on all the combined variables had a similar accuracy of 6.36%, indicating model created had reproducibility capability. Despite the outcome, the proposed method is not yet sufficiently strong for suspect identification. Further enhancement to improve its accuracy is required.

The prognostic role of controlling nutritional status and skeletal muscle mass in patients with hepatocellular carcinoma after curative treatment

Nutritional status and skeletal muscle mass affect the prognosis of hepatocellular carcinoma (HCC). Nutritional status is closely associated with skeletal muscle mass. Here, we investigate the effect of controlling preoperative nutritional status and skeletal muscle mass on the prognosis of HCC after curative treatment. This retrospective analysis contained 181 patients who received curative treatment of HCC including liver resection or radiofrequency ablation (RFA) therapy. Nutritional status and skeletal muscle mass were evaluated prior to therapy using the controlling nutritional status (CONUT) score and psoas muscle mass index (PMI), respectively. Associations of predictor variables with overall survival (OS) and progression-free survival (PFS) were determined using Cox proportional hazards regression and CHAID decision tree algorithm analysis. A total of 111 patients (61.3%) were determined to be of poor nutritional status and 100 patients (55.2%) had muscle mass depletion. Patients with PS 0, Barcelona clinic liver cancer (BCLC) stage 0, low CONUT score, and high PMI showed significantly better OS than those with PS 1, BCLC stage A, high CONUT score, and low PMI. Multivariate analysis indicated that a high CONUT score [hazard ratio (HR) 4.130; 95% confidence interval (CI), 1.713-9.958; P &lt; 0.01) and low PMI (HR 4.625; 95% CI, 1.704-12.549; P &lt; 0.01) found to be useful for predicting OS in patients after curative treatment of HCC. Regarding PFS, a significant predictor was only tumor numbers in univariate analysis (HR 2.147; 95% CI, 1.350-3.414; P = 0.001). In decision tree analysis, the mortality rate was 28.8%, 12.5%, and 1.9% in patients with a high CONUT score, with a low CONUT score-low PMI, or with a low CONUT score-high PMI, respectively. The combined CONUT score and PMI were found to be independent predictors of OS in HCC patients after liver resection or RFA.

Which Correlates Best Predict Physical Fitness Of Portuguese Children And Adolescents? A CHAID Decision Tree Analysis

Which Correlates Best Predict Physical Fitness Of Portuguese Children And Adolescents? A CHAID Decision Tree Analysis

The Algorithm for Using Transit-time Flow Measurement and High-resolution Epicardial Ultrasound for Intraoperative Grafts Assessment during Coronary Artery Bypass Surgery.

Aim To develop an algorithm for using ultrasonic flowmetry (USF) and epicardial ultrasonic scanning (EpiUSS) for intraoperative assessment of anatomic and functional viability of conduits.Material and methods For viability assessment of 460 coronary grafts in 150 patients who were operated at the Bakulev National Medical Research Center for Cardiovascular Surgery (2018-2021 г.), markers of graft failure were analyzed using the USF and EpiUSS data confirmed by results of graft angiography. According to RОС analysis, the Qmean and PI values indicative of the graft failure were determined. A CHAID decision tree was developed for assessing the prognostic significance of the analyzed parameters. Based on this prognostic model, an algorithm was developed for intraoperative diagnosis of anatomic and functional graft viability during coronary bypass surgery.Results The Qmean ≤20.5 ml/min values were associated with an increased relative risk (RR) of detecting graft failure (RR, 8.2; 95 % confidence interval, CI, 4.4-15.2). The developed model shows a high accuracy of predicting the graft failure (AUC = 0.906±0.03). The RR of graft failure at PI ≥2.65 was 3.3 (95 % CI, 2.17-5.08). The prognostic model for PI (AUC = 0.745±0.042) was sufficiently accurate with respect of possible graft failure. Nodes of high and low risk for graft failure were determined in the developed decision tree. The obtained model was characterized by high sensitivity and specificity (100 and 84.3 %, respectively).Conclusion The combined use of USF and EpiUSS allows a highly accurate assessment of both morphological and functional characteristics of graft flow. The developed algorithm for the intraoperative diagnosis of anatomic and functional graft viability can be recommended for clinical use.

A study on ski groups size and their relationship to the risk of injury

This paper addresses a novel topic in ski injury research - how a ski group size indicates the risk of ski injury. There is evidence in research literature that people ski in groups. However, the relationship between group size and the risk of injury has remained unexplored. Based on ski lift entrance data, we use the density-based clustering algorithm OPTICS to identify groups of skiers and discuss the advantages of using this algorithm. We show that the ski group size can be used to improve the identification of skiers who experience ski injury. The results of the identification of ski groups at Mt. Kopaonik Ski Resort in Serbia show that skiing alone is most susceptible to ski injury, while skiing in couples or in bigger groups reduces the risk of injury by 46%. In addition, it is confirmed that ski injuries are an early failure event phenomena. Based on the CHAID decision tree analysis, spending a small amount of time at the ski resort and skiing alone are associated with the: 6 times greater ski injury risk than the average ski injury risk.