Topic modeling and clustering in the trace data-driven analysis of job demands among teachers.

Psychosocial work environment characteristics like job demands have traditionally been studied using survey data. We propose an alternative approach utilizing work related trace data collected from the information systems that employees use to achieve organizational goals. We analyze the job demands of teachers from two universities of applied sciences using trace data collected from the educational online platform Moodle over a period of 90 weeks. The data contain pairs of targets and actions (like message_sent) performed by teachers on Moodle. The timestamps of the target-action pairs allow us to study the dynamic nature of job demands, which is not possible by using periodically collected survey data. We show how trace data can be used to analyze processes related to job demands using data-driven approaches. We have identified topics, themes, temporal processes, and employee clusters from Moodle data representing the work tasks of teachers. The information obtained is action-oriented, context-specific, and dynamic, meeting the current needs for information about changing working life. The approach we have provided could be widely utilized in organizations as well as in research on occupational wellbeing. It is useful in identifying targets for intervention and it could be expanded to include prediction models on different outcomes.

Habitat functionality: Integrating environmental and geographic space in niche modeling for conservation planning.

Niche modeling is typically used to assess the effects of anthropogenic land use and climate change on species distributions and to inform spatial conservation planning. These models focus on the suitability of local biotic and abiotic conditions for a species in environmental space (E-space). Although movements also affect species occurrence, efforts to formally integrate geographic space (G-space) into niche modeling have been hindered by the lack of comprehensive theoretical frameworks. We propose the "functional habitat" framework to define areas that are simultaneously of high quality in E-space, and functionally connected to other suitable habitats in G-space. Originating in metapopulation ecology, approaches have been developed to assess the amount of suitable connected habitats, based on the proximity between pairs of locations. Using network theory, which operates in topological space (T-space, defined by a network), we extended these metapopulation approaches to integrate movement constraints in G-space with niche modeling in E-space. We demonstrate the functional habitat framework using empirical data (GPS tracking and population monitoring) throughout the European wild mountain reindeer (Rangifer t. tarandus) distribution range. We show that functional habitat outperforms traditional suitability in explaining the species' distribution. This approach integrates effects from habitat loss and fragmentation for spatial conservation planning, and avoids overemphasizing small, inaccessible areas with locally suitable habitats. The functional habitat framework formally integrates biotic, abiotic, and movement constraints in niche modeling using network theory, thus opening a wide range of applications in spatial conservation planning.

Hospital physicians´ working hour characteristics and sleep quality: a cross-sectional analysis of realized working hour and survey data.

BACKGROUND: Hospital physicians' work includes on-call duties to provide 24/7 health care. Previous studies using self-reported survey data have associated long working hours and on-call work with sleep difficulties. To reduce recall bias, we complemented survey data with payroll-based objective data to study whether hospital physicians' realized working hours are associated with sleep. METHODS: The study was nested within the Finnish Public Sector study. We used survey data on 728 hospital physicians (mean age 43.4 years, 62% females) from 2015 linked to realized daily working hour data from 3 months preceding the survey. The associations of working hour characteristics with sleep quantity and quality were studied with multinomial logistic regression analysis adjusted for demographics, overall stressfulness of life situation, control over scheduling of shifts, and hospital district. RESULTS: One fourth (26%) of the participants reported short (≤6.5 h) average sleep duration. Frequent night work (> 6 shifts/91 days) was associated with short sleep (OR 1.87 95%CI 1.23-2.83) compared to no night work. Approximately one third (32%) of the physicians reported insufficient sleep. Physicians with long weekly working hours (> 48 hours) had higher odds for insufficient sleep (OR 1.78 95%CI 1.15-2.76) than physicians with short weekly working hours (< 40 hours). Insufficient sleep was also associated with frequent on-call duties (> 12 shifts/3 months OR 2.00 95%CI 1.08-3.72), frequent night work (OR 1.60 95%CI 1.09-2.37), and frequent short shift intervals (≤11 hours; > 12 times/3 months OR 1.65 95%CI 1.01-2.69) compared to not having these working hour characteristics. Nearly half of the physicians (48%) reported at least one sleep difficulty at least two times a week and frequent night work increased odds for difficulties in initiating sleep (OR 2.43 95%CI 1.04-5.69). Otherwise sleep difficulties were not associated with the studied working hour characteristics. CONCLUSION: We used realized working hour data to strengthen the evidence on on-call work and sleep quality and our results advice to limit the frequency of night work, on-call shifts, short shift intervals and long weekly working hours to promote hospital physicians' sufficient sleep.

Quantifying daily rhythms with non-negative matrix factorization applied to mobile phone data.

Human activities follow daily, weekly, and seasonal rhythms. The emergence of these rhythms is related to physiology and natural cycles as well as social constructs. The human body and its biological functions undergo near 24-h rhythms (circadian rhythms). While their frequencies are similar across people, their phases differ. In the chronobiology literature, people are categorized into morning-type, evening-type, and intermediate-type groups called chronotypes based on their tendency to sleep at different times of day. Typically, this typology builds on carefully designed questionnaires or manually crafted features of time series data on people's activity. Here, we introduce a method where time-stamped data from smartphones are decomposed into components using non-negative matrix factorization. The method does not require any predetermined assumptions about the typical times of sleep or activity: the results are fully context-dependent and determined by the most prominent features of the activity data. We demonstrate our method by applying it to a dataset of mobile phone screen usage logs of 400 university students, collected over a year. We find four emergent temporal components: morning activity, night activity, evening activity and activity at noon. Individual behavior can be reduced to weights on these four components. We do not observe any clear categories of people based on the weights, but individuals are rather placed on a continuous spectrum according to the timings of their phone activities. High weights for the morning and night components strongly correlate with sleep and wake-up times. Our work points towards a data-driven way of characterizing people based on their full daily and weekly rhythms of activity and behavior, instead of only focusing on the timing of their sleeping periods.

A bag-of-paths framework for network data analysis.

This work develops a generic framework, called the bag-of-paths (BoP), for link and network data analysis. The central idea is to assign a probability distribution on the set of all paths in a network. More precisely, a Gibbs-Boltzmann distribution is defined over a bag of paths in a network, that is, on a representation that considers all paths independently. We show that, under this distribution, the probability of drawing a path connecting two nodes can easily be computed in closed form by simple matrix inversion. This probability captures a notion of relatedness, or more precisely accessibility, between nodes of the graph: two nodes are considered as highly related when they are connected by many, preferably low-cost, paths. As an application, two families of distances between nodes are derived from the BoP probabilities. Interestingly, the second distance family interpolates between the shortest-path distance and the commute-cost distance. In addition, it extends the Bellman-Ford formula for computing the shortest-path distance in order to integrate sub-optimal paths (exploration) by simply replacing the minimum operator by the soft minimum operator. Experimental results on semi-supervised classification tasks show that both of the new distance families are competitive with other state-of-the-art approaches. In addition to the distance measures studied in this paper, the bag-of-paths framework enables straightforward computation of many other relevant network measures.

Two betweenness centrality measures based on Randomized Shortest Paths.

This paper introduces two new closely related betweenness centrality measures based on the Randomized Shortest Paths (RSP) framework, which fill a gap between traditional network centrality measures based on shortest paths and more recent methods considering random walks or current flows. The framework defines Boltzmann probability distributions over paths of the network which focus on the shortest paths, but also take into account longer paths depending on an inverse temperature parameter. RSP's have previously proven to be useful in defining distance measures on networks. In this work we study their utility in quantifying the importance of the nodes of a network. The proposed RSP betweenness centralities combine, in an optimal way, the ideas of using the shortest and purely random paths for analysing the roles of network nodes, avoiding issues involving these two paradigms. We present the derivations of these measures and how they can be computed in an efficient way. In addition, we show with real world examples the potential of the RSP betweenness centralities in identifying interesting nodes of a network that more traditional methods might fail to notice.

Predicting the continuum between corridors and barriers to animal movements using Step Selection Functions and Randomized Shortest Paths.

The loss, fragmentation and degradation of habitat everywhere on Earth prompts increasing attention to identifying landscape features that support animal movement (corridors) or impedes it (barriers). Most algorithms used to predict corridors assume that animals move through preferred habitat either optimally (e.g. least cost path) or as random walkers (e.g. current models), but neither extreme is realistic. We propose that corridors and barriers are two sides of the same coin and that animals experience landscapes as spatiotemporally dynamic corridor-barrier continua connecting (separating) functional areas where individuals fulfil specific ecological processes. Based on this conceptual framework, we propose a novel methodological approach that uses high-resolution individual-based movement data to predict corridor-barrier continua with increased realism. Our approach consists of two innovations. First, we use step selection functions (SSF) to predict friction maps quantifying corridor-barrier continua for tactical steps between consecutive locations. Secondly, we introduce to movement ecology the randomized shortest path algorithm (RSP) which operates on friction maps to predict the corridor-barrier continuum for strategic movements between functional areas. By modulating the parameter Ѳ, which controls the trade-off between exploration and optimal exploitation of the environment, RSP bridges the gap between algorithms assuming optimal movements (when Ѳ approaches infinity, RSP is equivalent to LCP) or random walk (when Ѳ → 0, RSP → current models). Using this approach, we identify migration corridors for GPS-monitored wild reindeer (Rangifer t. tarandus) in Norway. We demonstrate that reindeer movement is best predicted by an intermediate value of Ѳ, indicative of a movement trade-off between optimization and exploration. Model calibration allows identification of a corridor-barrier continuum that closely fits empirical data and demonstrates that RSP outperforms models that assume either optimality or random walk. The proposed approach models the multiscale cognitive maps by which animals likely navigate real landscapes and generalizes the most common algorithms for identifying corridors. Because suboptimal, but non-random, movement strategies are likely widespread, our approach has the potential to predict more realistic corridor-barrier continua for a wide range of species.