WEBA Dataset
WEBA dataset: a facilitator for the study of work content effect on stress, workload perception and individual performance in real-life working conditions
The story behind the dataset
Our Operator 4.0 research group has long focused on exploring human-centric approaches in industrial environments, emphasizing human factors, process analysis, and performance optimization, enhancing both productivity and their mental/physical well-being. One recent study centered on how work content, such as the physical, mental and temporal demands of tasks, can be captured with physiological signals of the industrial operators, like heart rate variability [1].
During our research, it is occurred that we did not find any evidentiary evidence that truly reflects the effect of work content, especially in real-life working conditions. Other laboratory experiment or observational studies either lack of the work content/work context conceptualization, or failed to separate the corporate effects of them. These obstacles hinder deriving a meaningful conclusion about how work content factors affect the physiological stress and performance of the workers.
With that concern in mind, fortunately, we found a coffee shop in Budapest, the capital of Hungary, that contains all the similar characterstics to an industrial environment with repetitive cycles of standardized and descrete incoming tasks, but has an unique setup that support controlling and maintaining the other work context, work environment factors. By that way, the work content become the sole and main stressors, which exert effects on the working baristas, and mandates their stress, workload perception, performance. We realized that our knowlegde from the industrial context can be utilized in this context, as a living laboratory.
[1] Tran, T. A., Péntek, M., Motahari-Nezhad, H., Abonyi, J., Kovács, L., Gulácsi, L., ... & Ruppert, T. (2023). Heart Rate Variability Measurement to Assess Acute Work-Content-Related Stress of Workers in Industrial Manufacturing Environment—A Systematic Scoping Review. IEEE Transactions on Systems, Man, and Cybernetics: Systems.
Work content Effect on BAristas (WEBA) dataset
Figure 1. The coffee shop where the data collection was carried out
Considering the work characteristics of the baristas, the work content can be categorized into mental workload as the difficulty of the incoming ordered drinks, physical workload as the frequency and intensity of the physical activities during making and serving drinks, and temporal workload as the required work pace. The work context that also poses stress on the barista includes the interactions with customers, and non-work-content-related activities that are not associated with the number of ordered drinks (e.g., floor cleaning, interaction with the managers). The work environment contains factors that reflect the physical environment, such as working area, ergonomic arrangement, lighting condition, noise, and temperature, etc.
The coffee shop’s confined environment and compact layout limited the number of employees to only one barista working at a time, naturally isolating work content as the primary factor influencing stress and performance, similar to a controlled environment in laboratory setup.
The work content includes mainly the coffee-making process itself, which somehow consists of repetive tasks following standard steps in the drink’s recipes, resembled the standardized workflows - all akin to industrial operators carrying out repetitive work cycles. It is conceivable that the work content of the barista requires mainly effort from the physical aspect, such as taking orders, making drinks, plating, serving, etc. From the mental aspect, it is assumed that the difficulty of each order creates a different impact on the perceived workload of individual, mainly depending on their work experience. Regarding the temporal aspect, the intensity of incoming orders with the quantity of each order will create time pressure on the barista.
Figure 2. The work process is the combination of the “work content” and “work context” within the “work environment”, while the result is the “output” in quantity and monetary value, along with the “work content effect” on the barista.
To comply with the General Data Protection Regulation (GDPR), no video or image was captured. Instead, sensors were installed with fixed locations within the working environment to facilitate activity recognition of surrounding events, while wearables were equipped on the barista to recognize the performed activities and record physiological signal. This approach allowed us to monitor baristas’ physical movements, heart rates, and perceived workloads under varying task complexities and paces. For example, we recorded shifts with different task combinations, observing how these variables influenced stress and performance metrics. To assess the amount of work content within a working shift, objective and subjective measures were taken once the barista finished that shift, with the produced revenues and order timestamps extracted from the ordering system, along with the quantity and type of sold drinks and cakes.
Figure 3. One hour of raw data between 3 pm and 4 pm from the “20240223_evening” shift conducted by the fourth participant. The produced revenue is 46.725 HUF. This work shift is perceived by the barista as an 18/8 positive/negative score on the I-PANAS-SF scale
Besides of Heart Rate (HR) collection during work shifts, a data collection is performed with two days of 24-hour continuous HR measurement with each human participant, to provide a comparable baseline to the working hours during the measured work shifts.
WEBA dataset contains the physical work activities and heart rates of five baristas in 55 shifts with different work content combinations. An example of one hour of raw data from an evening shift conducted by the fourth barista can be seen in below figure. The collected data from Tapo sensors can be grouped into customer-oriented events (e.g., opening/closing the front floor, passing the door, placing an order, going upstairs) or work-oriented events (e.g., moving in/out of Zone_1 and Zone_2, open/close the fridge, the cake display, the freezer). The acceleration data were plotted separately for the body and the hand. In the last subplot, the HR value is visualized, with the baseline value on weekdays and weekends of the same participant shown as a benchmark for easy recognition.
The WEBA dataset contributes to further developing an understanding of the work content effect on labor performance and well-being. This work contributes a piece of evidence for the use of physiological parameters in reflecting the acute work-content-related stress (AWCRS), where the possibility of using HR and acceleration signals as indicators for AWCRS can be diagnosed, and other impact of the work content on granular emotion and performance can be analyzed.
Data access
The dataset can be accessed in figshare at:
https://doi.org/10.6084/m9.figshare.27186516.v3
When using the WEBA dataset, please refer to this paper:
Tran, T. A., Eigner, G., Abonyi, J., & Ruppert, T. (2025). WEBA dataset as the Reflection of Work content effect on Workload perception in Real life Working conditions. Scientific Data, 12(1), 11.
Aknowledgement
This work has been implemented by the TKP2021-NVA-10 project with the support provided by the Ministry of Culture and Innovation of Hungary from the National Research, Development and Innovation Fund, financed under the 2021 Thematic Excellence Programme funding scheme. The work of Tamás Ruppert is supported by the János Bolyai Research Scholarship of the Hungarian Academy of Science. The WEBA dataset was collected with the support of the 2024-2.1.1-EKÖP University Research Scholarship Programme of the Ministry for Culture and Innovation, Hungary from the source of the National Research, Development, and Innovation Fund, Hungary.