FabLab SUPSI DACD - Machine and Training Usage Dataset

The Digital Fabrication and Open Innovation Laboratory (FabLab) - SUPSI located on the DACD campus of SUPSI is a facility that supports students and staff of the university in the physical production of digital projects in the fields of architecture, interior design, engineering, design, conservation and restoration, and microbiology.

The laboratory offers various machines for rapid prototyping, including: laser cutting machines, small and large format 3D printers, numerically controlled milling machines, vinyl cutters and an electronics workstation. In order to use the machines, it is necessary to take a vertical training course on the machine and the safety practices to be followed for its use. Authorised users can then book the equipment in 20-minute slots, either free of charge or for a fee, depending on the equipment. The booking of training courses and machine slots is managed through an open source platform called FabManager. The local instance can be accessed at fabmanager.supsi.ch. After registering via SSO, SUPSI users can log in to their management profile. On the platform, it is possible to view the details of each machine, finding technical data sheets and information about courses and available slots.

As the platform administrator, it is possible to access a user management panel and a set of APIs for analysing data related to the laboratory. From the backend, statistics can be easily viewed through an intuitive user interface, with the ability to filter data by user, date, gender, age, type and cost.

Dataset content and retrevial

Direct database interrogation via platform-provided Application Programming Interfaces (APIs) is employed.
Data are retrieved utilising a custom Python package, that is released via GitHub, resulting in clean tabular data formatted as JSON and comprising the following entries: users (number not public), machinery bookings (n=10639), course registrations (n=1477), a list of machines (n=34), and a list of courses (n=29). Due to the absence of publicly available APIs for projects retrieval, these records are excluded from this dataset. The specific field definitions applicable to each data category are comprehensively documented within the FabLab_DACD_SUPSI_UsageData_Schema.json document.

Data cleaning process

The raw data exported from the APIs is analysed and minimised to guarantee anonymisation. The analysis of the actual type of data is carried out below.

Machine data

The dataset pertaining to the machines within the digital fabrication laboratory of DACD contains information regarding all machinery, both operational and disabled. As some records relate to equipment not publicly visible, identified by ‘disabled’: true, these are removed from the published dataset to prevent sharing details concerning machinery unavailable to lab users. The dataset also includes a slug referencing each machine’s public-facing webpage; this is completed with the domain name to facilitate Linked Open Data connectivity. Creation dates of machines may serve as useful provenance information, indicating when equipment was added to the laboratory's inventory, only the date component is retained. Update dates are deemed unnecessary for potential analyses.

Machine descriptions and specifications currently utilise HTML formatting, which hinders readability; therefore, a cleaning process eliminates superfluous tags while preserving external links to maintain connections with relevant sources. Metadata appended at the end of each file includes: data and time of cleaning, Data Owner (the dataset’s owner), Data Steward (responsible for data cleansing), Data Curator (responsible for initial data collection), source system from which the data are extracted ("Data Exported from"), date and time of export (“Data Exported at”), and licensing terms.

Training data

The dataset pertaining to the training courses offered at the digital fabrication laboratory of DACD contains data relating to all available courses. These records include course names, slugs linking to corresponding webpages, a flag indicating public visibility, creation and modification dates, the number of places available per session, and descriptions formatted in HTML. Following similar principles as previously described, this dataset undergoes cleaning according to established procedures; the maximum participant capacity is retained due to its potential utility for analytical purposes without compromising data integrity. The metadata remain consistent with those detailed earlier.

Machine reservation data

This dataset compiles all machinery bookings made at the laboratory, including user and machine details. To protect privacy, only the following data elements are retained: creation date of the booking, user group (course of study), identifier and link to the booked machine, an indicator denoting cancellation status, and duration of the reservation in hours. The metadata adhere to the conventions established for previous datasets.

Training reservation data

This dataset mirrors the structure of its predecessor, incorporating various details pertaining to booked training courses. The following data elements are retained within this recordset: booking creation date, user group affiliation, course identification and URL link, and an indicator denoting cancellation status.

Folder structure

The directory comprises three files at the primary level.

FabLab_DACD_SUPSI_UsageData_Schema.json contains the dataset schema formulated according to JSON Schema Draft 7, incorporating supplementary details such as version information, authorial attribution, affiliation, creation and update dates, alongside requisite schema elements. A descriptive account of each field is included, accompanied by data type specifications and formatting particulars where applicable.

FabLab_DACD_SUPSI_UsageData_Metadata_Export_02_12_2025.json exclusively houses metadata—also incorporated within the dataset itself—providing specific information pertaining to dates of data merging, data ownership, stewardship details (including ORCID identifiers), curation responsibilities (with corresponding ORCID identifier), source system for exported data, licensing terms, DOI assignment, and machine-specific export/cleaning timestamps for both machinery and training data.

Finally, FabLab_DACD_SUPSI_UsageData_Export_02_12_2025.json represents the dataset itself, incorporating metadata as previously described alongside the substantive data arranged into categories: machines, trainings, reservations_machine, and reservations_training; structural details for each category are delineated within the schema document.

Ethics risks

It is possible to distinguish between the use of the machines by groups of students, which could limit the sample of users who may have used the machine at that time to around 40 people. This could inadvertently facilitate correlation between specific users and the equipment use on particular days; however, this association is not deemed inherently detrimental given the complexity involved in definitively identifying individual user identities and the actual information obtained, thus precluding any required remedial action.

Sharing machine usage data allows for assessment of laboratory presence rates, a form of monitoring information that holds potential for misuse. However, as the facility operates within a public setting, no mitigating actions are considered necessary; furthermore, this information is crucial for dataset analysis.

Within the training course descriptions, teaching staff names and surnames frequently appear. No specific measures are implemented in response to this occurrence, given their existing availability on the laboratory’s publicly accessible website.

Technical risks

Due to its considerable size, manual assessment of this dataset proves challenging; therefore, minimisation and anonymisation were wholly executed via Python scripting. While there remains a theoretical possibility of inadvertently sharing non-public data through this automated process, the robust JSON schema structure significantly mitigates such risk.

The entire dataset has been restructured according to a defined logic, simplified within the accompanying scheme file; however, unfamiliarity with this new architecture may impede analytical efforts.