Machine Learning-Based Predictions of Customers’ Decisions in Car Insurance

Predicting customer decisions allows companies to obtain higher profits due to better resource management. The accuracy of those predictions can be currently boosted by the application of machine learning algorithms.

We propose a new method to predict a car driver’s decision about taking a replacement car after a vehicle accident happens. We use feature engineering to create attributes of high significance. The generated attributes are related to time (e.g., school holidays), place of collision (e.g., distance from home), time and conditions (e.g., weather), vehicles (e.g., vehicle value), addresses of both the victim and the perpetrator. Feature engineering involves external sources of data.

Five machine learning methods of classification are considered: decision trees, multi-layer perceptrons, AdaBoost, logistic regression and gradient boosting. Algorithms are tested on real data from a Polish insurance company. Over 80% accuracy of prediction is achieved. Significance of the attributes is calculated using the linear vector quantization method.

Presented work shows the applicability of machine learning in the car insurance market.     

Comparison of visual fatigue caused by head-mounted display for virtual reality and two-dimensional display using objective and subjective evaluation

This study aimed to evaluate objective and subjective visual fatigue experienced before and after performing a visual task while using a head-mounted display for virtual reality (VR-HMD) and two-dimensional (2D) display. Binocular fusion maintenance (BFM) was measured using a binocular open-view Shack–Hartmann wavefront aberrometer equipped with liquid crystal shutters. Twelve healthy subjects performed the BFM test and completed a questionnaire regarding subjective symptoms before and after performing a visual task that induces low visually induced motion sickness (VIMS). BFM (p = .87) and total subjective eye symptom scores (p = .38) were not significantly different between both groups, although these values were significantly lower after the visual task than before the task within both groups (p < .05). These findings suggest that visual fatigue after using a VR-HMD is not significantly different from that after using a 2D display in the presence of low-VIMS VR content.

Practitioner summary: Objective and subjective evaluation of visual fatigue were not significantly different with the use of a head-mounted display for virtual reality (VR-HMD) and two-dimensional display. These results should be valuable not only to engineers developing VR content but also to researchers involved in the evaluation of visual fatigue using VR-HMD.

Abbreviations: VR: virtual reality; VR-HMD: head-mounted display for virtual reality; BFM: binocular fusion maintenance; BWFA: binocular open-view Shack–Hartmann wavefront aberrometer     

Forecasting Ranking in Harness Racing Using Probabilities Induced by Expected Positions

Ranked events are pivotal in many important AI-applications such as Question Answering and recommendations systems. This paper studies ranked events in the setting of harness racing.

For each horse there exists a probability distribution over its possible rankings. In the paper, it is shown that a set of expected positions (and more generally, higher moments) for the horses induces this probability distribution.

The main contribution of the paper is a method, which extracts this induced probability distribution from a set of expected positions. An algorithm is proposed where the extraction of the induced distribution is given by the estimated expectations. MATLAB code is provided for the methodology.

This approach gives freedom to model the horses in many different ways without the restrictions imposed by for instance logistic regression. To illustrate this point, we employ a neural network and ordinary ridge regression.

The method is applied to predicting the distribution of the finishing positions for horses in harness racing. It outperforms both multinomial logistic regression and the market odds.

The ease of use combined with fine results from the suggested approach constitutes a relevant addition to the increasingly important field of ranked events.     

Introducing ergonomics requirements in the eco-design of energy-related products from users’ behaviour approach

Ergonomics has been a very important activity in the design process. However, ergonomics rarely includes the environmental requirements into the design of products. The article proposes and presents the Eco-Ergo model through its application to a real-world product, a washing machine, to allow designers and ergonomists to establish product design requirements in order to minimise environmental impacts related to user–product interaction during the use stage. This model uses a visual language of representation, Blueprinting-based, that helps designers explore problems they have not previously considered during the market research when a wide variety of products with different interaction elements is analysed. The application of this model allows direct efforts and attention on the user analysis phase in the most influential user’s actions on the environmental performance of energy-related products during use, establishing ergonomics requirements related to users behaviour at the initial design phase.

Practitioner summary: This study provides a proposal to incorporate ergonomics into the practice of eco-design through the use of human factors in the establishment of initial eco-design requirements. This blueprint-based model combines an empirical and theoretical approach, based on the product test developed by designers, ergonomists and environmentalists.

Abbreviations: CO2: carbon dioxide; DBIM: design behaviour intervention model; DfSB: design for sustainable behaviour; DwI Method: design with intent method; LCA: life cycle assessment; LCD: liquid cristal display; MJ: megajoule; MTM: methods-time measurement; PSS: product-service system; RPM: revolutions per minute     

Retention of a standard operating procedure under the influence of social stress and refresher training in a simulated process control task

In a simulated process control task, we investigated the effects of refresher training and acute social stress on performing a standard operating procedure (SOP) containing a production and monitoring task and knowledge recall after a retention interval of two weeks. In a 2?×?2 between-group design (Factor 1: induced social stress, Factor 2: refresher training), 76 engineering students performed an SOP at t1 in week 1 and at t2 in week 3. A MANOVA in week three (t2) indicated a main effect of the refresher training for the SOP execution containing a production and a monitoring task and an impairing effect of stress on the monitoring task. That means that after a retention interval, stress mainly affects the SOP’s monitoring task. An additional correlational analysis showed that knowledge test performance is negatively associated with cortisol level and that retentivity is a strong predictor for knowledge test performance and production task performance, too.

Practitioner Summary: We investigated effects of social stress and refresher training on performing a standard operating procedure (SOP) after a retention interval of two weeks. The impact of social stress reduced the monitoring task performance as part of the SOP, but not the production outcome. Without refresher training, performance is significantly worse.

Abbreviation: SOP: Standard Operating Procedure; MANOVA: Multivariate Analysis of Variance; CSB: Chemical Safety and Hazard Investigation Board; TSST: Trierer Social Stress Test; P-TSST: Placebo Trierer Social Stress Test; WaTrSim: Water Treatment Simulation; HPA axis: hypothalamic pituitary adrenal axis; WIT-2: Wilde Intelligenztest – 2; SPSS: Statistical Package for the Social Sciences; ANOVA: Analysis of Variance.     

Posture normalisation of 3D body scans

For product developers that design near-body products, virtual mannequins that represent realistic body shapes, are valuable tools. With statistical shape modelling, the variability of such body shapes can be described. Shape variation captured by statistical shape models (SSMs) is often polluted by posture variations, leading to less compact models. In this paper, we propose a framework that has low computational complexity to build a posture invariant SSM, by capturing and correcting the posture of an instance. The posture-normalised SSM is shown to be substantially more compact than the non-posture-normalised SSM.

Practitioner summary: Statistical shape modelling is a technique to map out the variability of (body) shapes. This variability is often polluted by variations in posture. In this paper, we propose a framework to build a posture invariant statistical shape model.

Abbreviations: SSM: statistical shape model; 1D: one-dimensional; 3D: three-dimensional; DHM: digital human model; LBS: linear blend skinning; PCA: princial component analysis; PC: principal component; TTR: thumb tip reach.     

The impact of external facial features on the construction of facial composites

Witnesses may construct a composite face of a perpetrator using a computerised interface. Police practitioners guide witnesses through this unusual process, the goal being to produce an identifiable image. However, any changes a perpetrator makes to their external facial-features may interfere with this process. In Experiment 1, participants constructed a composite using a holistic interface one day after target encoding. Target faces were unaltered, or had altered external-features: (i) changed hair, (ii) external-features removed or (iii) naturally-concealed external-features (hair, ears, face-shape occluded by a hooded top). These manipulations produced composites with more error-prone internal-features: participants’ familiar with a target’s unaltered appearance less often provided a correct name. Experiment 2 applied external-feature alterations to composites of unaltered targets; although whole-face composites contained less error-prone internal-features, identification was impaired. Experiment 3 replicated negative effects of changing target hair on construction and tested a practical solution: selectively concealing hair and eyes improved identification.

Practitioner Summary: The research indicates that when a target identity disguises or changes hair, this can lead to a witness (or victim) constructing a composite that is less readily identified. We assess a practical method to overcome this forensic issue.

Abbreviation: GEE: Generalized Estimating Equations     

Role network measures to assess healthcare team adaptation to complex situations: the case of venous thromboembolism prophylaxis

Hospitals are complex environments that rely on clinicians working together to provide appropriate care to patients. These clinical teams adapt their interactions to meet changing situational needs. Venous thromboembolism (VTE) prophylaxis is a complex process that occurs throughout a patient’s hospitalisation, presenting five stages with different levels of complexity: admission, interruption, re-initiation, initiation, and transfer. The objective of our study is to understand how the VTE prophylaxis team adapts as the complexity in the process changes; we do this by using social network analysis (SNA) measures. We interviewed 45 clinicians representing 9 different cases, creating 43 role networks. The role networks were analysed using SNA measures to understand team changes between low and high complexity stages. When comparing low and high complexity stages, we found two team adaptation mechanisms: (1) relative increase in the number of people, team activities, and interactions within the team, or (2) relative increase in discussion among the team, reflected by an increase in reciprocity.

 

Practitioner Summary: The reason for this study was to quantify team adaptation to complexity in a process using social network analysis (SNA). The VTE prophylaxis team adapted to complexity by two different mechanisms, by increasing the roles, activities, and interactions among the team or by increasing the two-way communication and discussion throughout the team. We demonstrated the ability for SNA to identify adaptation within a team.     

Sex differences in upper limb 3D joint contributions during a lifting task

Sex-related differences in work technique may contribute to increasing the risk of musculoskeletal joint disorders among women. In lifting tasks, sex differences have been reported for the trunk and lower limb, although women present a higher prevalence of shoulder disorders. We investigated sex differences in the upper limb technique during a lifting task. Trunk and upper limb kinematics were recorded in 27 women and 27 men lifting a box (6 or 12?kg) from hip to eye level. Work technique was quantified through the three-dimensional contribution of each joint to overall box height. The glenohumeral joint showed a higher contribution in women with a 6?kg box and wrist and elbow joints did with a 12?kg box, compared to men at either 6 or 12?kg. Sex differences occurred systematically above shoulder level. Our results argue for careful consideration of sex during ergonomic intervention, particularly during the overhead task.

Practitioner Summary: We investigated the sex-related differences in upper limb technique during lifting tasks. Results highlight a sex-specific kinematic strategy above the shoulder level on the glenohumeral joint and on the wrist and elbow joints. To help reduce women’s shoulder disorders in overhead task, ergonomic interventions should account for those differences.

Abbreviations: DoF: degree-of-freedom; WR/EL: wrist and elbow; GH: glenohumeral; SC/AC: sternoclavicular and acromioclavicular; TR/PE: pelvo-thoracic     

The effect of age on in-vivo spine stiffness,postures and discomfort responses during prolonged sitting exposures

Many industrialised working populations are ageing and prolonged sitting exposures are prevalent across occupational sectors. The purpose of this work was to determine the effect of age and sex on passive spine stiffness, postures and discomfort in response to seated work. A total of 34 participants were recruited, with 17 older adults with an average age of (standard deviation) 63.7 (±3.9) years and 17 younger adults aged 23.8 (±5.0) years. Participants were asked to sit continuously for 90?min while typing. Baseline passive spine stiffness was higher in older adults at 40% flexion compared to younger adults (p?=?.0233). Older adults sat in less normalised flexion, at 33.4% (±16.4) compared to 60.9% (±20.2) in the younger group (p?=?.0003). Discomfort was higher among older adults in the neck, right shoulder and middle back regions (p?<?.0086). An understanding of age-specific responses to workplace exposures is essential to determine whether age-specific interventions are warranted.

 

Practitioner summary: Older adults had higher passive spine stiffness and sat with less flexion during prolonged sitting. Discomfort was higher among older adults and occurred earlier in the simulation compared to younger participants, indicating that interventions, such as walking breaks may need to be implemented earlier during sitting for aged workers.     

Where to put the creases? Interactions between hand length,task, screen size,and folding method on the suitability of hand-held foldable display devices

Limited information is available regarding ergonomic foldable display device forms. This two-stage study involving young South Koreans (divided into three hand-length groups) was conducted to determine ergonomic forms for hand-held foldable display devices considering folding/unfolding comfort and preference. Stage I obtained the suitability of three screen sizes for five tasks. Stage II evaluated 14 different bi- and tri-folding methods considering screen size, folding direction, and folding time. The effects of hand length were all non-significant. Screen size preferences were task-dependent; small screens were preferred for making calls, and medium screens for web searching and gaming. Folding methods affected folding/unfolding comfort and preference; outward screen and Z-shape screen folding were the most preferred bi- and tri-fold concepts, respectively. Screen protection and access appeared to be competing factors in the user preference determination process. Foldable screen size and folding method should be determined by considering tasks, folding/unfolding comfort, and user preferences.

Practitioner summary: A 13.5?cm screen was preferred for making calls, whereas a 17.5?cm screen was best for web searching and gaming. An outward bi-fold screen concept with a 17.5?cm screen and Z-shape tri-fold screen concept with a 22.9?cm screen were preferred. Overall, the Z-shape concept was most preferred.

Abbreviations: SD: standard deviation; ANOVA: analysis of variance; H: Height; W: Width; T: Thickness     

Effects of backpack and double pack loads on postural stability

Measurement of postural stability is crucial for identifying predictors of performance, determining the efficacy of physical training and rehabilitation techniques and evaluating and preventing injuries, particularly for heavy load carriage in hikers, mountain search and rescue personnel and soldiers. This study investigated the effect of load distribution on postural stability in an upright stance using backpack and double pack loads under conflicting or impaired somatosensory, visual and vestibular conditions. The sensory organisation tests were conducted on 20 young adults before and after a 10-min level walking exercise. Young adults’ ability to use inputs from somatosensory and visual systems to maintain postural stability was significantly reduced following a 10-min walking exercise with a heavy backpack (30% of body weight), whereas no significant changes were observed for double pack carriage. Thus, the distribution of heavy loads to the front and back provides superior balance control compared with back-only loading.

Practitioner summary: This study investigated the effects of heavy (30% of body weight) load distribution on postural stability after a 10-min walking exercise. Backpack carriage significantly reduced postural stability, whereas there was no significant effect under double pack loads. Distribution of heavy loads on the front-and-back is desirable for superior balance control.     

Evaluating sociotechnical dynamics in a simulated remotely-piloted aircraft system: a layered dynamics approach

As coordination mechanisms change and technology failures occur, a sociotechnical system must reorganise itself across human and technological layers to maintain effectiveness. We present a study examining reorganisation across communication, controls and vehicle layers of a remotely-piloted aircraft system (RPAS) using a layered dynamics approach. Team members (pilot; navigator; photographer) performed 5 simulated RPAS missions using different operator configurations, including all-human and human-autonomy teams. Reorganization (operationally defined using entropy) time series measured the changing system reorganisation profiles under different operator configurations and following autonomy failures. Correlations between these reorganisation profiles and team effectiveness scores describe the manner in which the system had to be coordinated to maintain effectiveness under these changing conditions. Four unplanned autonomy failures were analysed to visualise system reorganisation following a technology failure. With its objective and real-time modelling and measurement capabilities, layered dynamics complements existing systems thinking tools for understanding sociotechnical complexity and enhancing system effectiveness.

Practitioner summary: A layered dynamics approach for understanding how a sociotechnical system dynamically reorganises itself is presented. The layered dynamics of RPAS were analysed under different operator configurations and following autonomy failures. Layered dynamics complements existing system-thinking tools for modelling sociotechnical system complexity and effectiveness.

Abbreviation: RPAS: remotely-piloted aircraft system; HIS: human-systems integration; EAST: event analysis of systemic teamwork; H1: hypothesis 1; H2: hypothesis 2; H3: hypothesis 3; CERTT-STE: cognitive engineering research on team tasks--synthetic task environment; AVO: air vehicle operator; PLO: payload operator; DEMPC: data exploitation, mission planning, and communications; ACT-R: adaptive control of thought-rational; sec: seconds; ANOVA: analysis of variance     

The effect of a passive trunk exoskeleton on metabolic costs during lifting and walking

The objective of this study was to assess how wearing a passive trunk exoskeleton affects metabolic costs, movement strategy and muscle activation during repetitive lifting and walking. We measured energy expenditure, kinematics and muscle activity in 11 healthy men during 5?min of repetitive lifting and 5?min of walking with and without exoskeleton. Wearing the exoskeleton during lifting, metabolic costs decreased as much as 17%. In conjunction, participants tended to move through a smaller range of motion, reducing mechanical work generation. Walking with the exoskeleton, metabolic costs increased up to 17%. Participants walked somewhat slower with shortened steps while abdominal muscle activity slightly increased when wearing the exoskeleton. Wearing an exoskeleton during lifting decreased metabolic costs and hence may reduce the development of fatigue and low back pain risk. During walking metabolic costs increased, stressing the need for a device that allows disengagement of support depending on activities performed.

Practitioner summary: Physiological strain is an important risk factor for low back pain. We observed that an exoskeleton reduced metabolic costs during lifting, but had an opposite effect while walking. Therefore, exoskeletons may be of benefit for lifting by decreasing physiological strain but should allow disengagement of support when switching between tasks.

Abbreviations: COM: centre of mass; EMG: electromyography; LBP: low back pain; MVC: maximum voluntary isometric contraction; NIOSH: National Institute for Occupational Safety and Health; PLAD: personal lift augmentation device; PWS: preferred walking speed without exoskeleton; PWSX: preferred walking speed with exoskeleton; ROM: range of motion; RER: respiratory exchange ratio; V ?O2max: maximum rate of oxygen consumption     

An experimental study on individual and group affect in multi-tasking teams

Research has shown that affect can influence human performance in various settings. This study aimed to explore the effects of individual and group affect on task perceptions and performance under different stress conditions (task demand and technology reliability) and team training methods in multi-tasking environments. Seventy-two participants participated in the experiment in two-person teams. The participants’ affect was measured using facial expression recognition technology. Affect aggregated mean and affect similarity were used as indicators of group affect. Self-reports of workload, teamwork satisfaction, trust in team, and trust in technology were collected. The results demonstrated that different stress conditions and training methods altered individual and group affect. Individual affect did not significantly correlate with the task perception measures, while group affect was related to performance in the team-oriented task but not the individual-oriented tasks. These findings suggested that group affect may have an important role in both multi-tasking and team coordination.

Practitioner summary: Affect can influence human performance in various settings. This study explored how affect influences task perceptions and performance for teams in multi-tasking environments. The results demonstrated that individual affect was not correlated with self-reported task perceptions. Group affect was related to team performance in a team-oriented task.     

Functional near-infrared spectroscopy in the evaluation of urban rail transit drivers’ mental workload under simulated driving conditions

The objective of this study is to investigate the potential of functional near-infrared spectroscopy (fNIRS) combined with heart rate variability indices, for the evaluation of the mental workload of urban rail transit drivers under simulated driving conditions, particularly during task engagement and disengagement. Experienced metro drivers wearing fNIRS monitoring systems were asked to drive for 90?min in a professional metro driving simulator. Workload stimulus tasks were added and an n-back task (n?=?3) was implemented to induce workload in the simulated driving experiment. Experimental results indicate that fNIRS are sensitive to mental workload and reliable for discriminating the degree of mental workload. Research findings demonstrate the feasibility and reliability of fNIRS as a tool for real-time evaluating and monitoring driver mental workload along with task factors from a perspective of brain activations during simulated or actual driving.

Practitioner Summary: This study provides evidence for the potential of functional near-infrared spectroscopy (fNIRS) for the evaluation of the mental workload of urban rail transit drivers under simulated driving conditions. The first fNIRS application to mental workload evaluation in the field of urban rail transportation helps companies develop reasonable shiftwork schedule and ensure operation safety.

Abbreviations: fNIRS: functional near-infrared spectroscopy; oxy-Hb: Oxy-hemoglobin; NASA-TLX: National Aeronautics and Space Administration Task Load Index; EEG: electroencephalogram; ECG: electrocardiogram; HRV: variability; LF: low-frequency power; HF: high-frequency power; PFC: prefrontal cortex; NIRS: near-infrared spectroscopy; DWT: discrete wavelet transform; EMG: electromyography; DT: determination test; TP: total power; LF_norm: standardized LF; HF_norm: standardized HF; VLF: very low frequency; deoxy-Hb: deoxy-hemoglobin.     

Seven HCI Grand Challenges

This article aims to investigate the Grand Challenges which arise in the current and emerging landscape of rapid technological evolution towards more intelligent interactive technologies, coupled with increased and widened societal needs, as well as individual and collective expectations that HCI, as a discipline, is called upon to address. A perspective oriented to humane and social values is adopted, formulating the challenges in terms of the impact of emerging intelligent interactive technologies on human life both at the individual and societal levels. Seven Grand Challenges are identified and presented in this article: Human-Technology Symbiosis; Human-Environment Interactions; Ethics, Privacy and Security; Well-being, Health and Eudaimonia; Accessibility and Universal Access; Learning and Creativity; and Social Organization and Democracy. Although not exhaustive, they summarize the views and research priorities of an international interdisciplinary group of experts, reflecting different scientific perspectives, methodological approaches and application domains. Each identified Grand Challenge is analyzed in terms of: concept and problem definition; main research issues involved and state of the art; and associated emerging requirements.

BACKGROUND

This article presents the results of the collective effort of a group of 32 experts involved in the community of the Human Computer Interaction International (HCII) Conference series. The group’s collaboration started in early 2018 with the collection of opinions from all group members, each asked to independently list and describe five HCI grand challenges. During a one-day meeting held on the 20th July 2018 in the context of the HCI International 2018 Conference in Las Vegas, USA, the identified topics were debated and challenges were formulated in terms of the impact of emerging intelligent interactive technologies on human life both at the individual and societal levels. Further analysis and consolidation led to a set of seven Grand Challenges presented herein. This activity was organized and supported by the HCII Conference series.     

Detecting Premature Ventricular Contraction by Using Regulated Discriminant Analysis with Very Sparse Training Data

Pathological electrocardiogram is often used to diagnose abnormal cardiac disorders where accurate classification of the cardiac beat types is crucial for timely diagnosis of dangerous conditions. However, accurate, timely, and precise detection of arrhythmia-types like premature ventricular contraction is very challenging as these signals are multiform, i.e. a reliable detection of these requires expert annotations.

In this paper, a multivariate statistical classifier that is able to detect premature ventricular contraction beats is presented. This novel classifier can be trained with a very sparse amount of expert annotated data. To enable this, the dimensionality of the feature vector is kept very low, it uses strong designed features and a regularization mechanism. This approach is compared to other classifiers by using the MIT-BIH arrhythmia database. It has been found that the average accuracy, specificity, and sensitivity are above 96%, which is superior given the sparse amount of training data.