Improving time use measurement with personal big data collection - the experience of the European Big Data Hackathon 2019

Mattia Zeni, Ivano Bison, Britta Gauckler, Fernando Reis, Fausto Giunchiglia

Abstract

This article assesses the experience with i-Log at the European Big Data Hackathon 2019, a satellite event of the New Techniques and Technologies for Statistics (NTTS) conference, organised by Eurostat. i-Log is a system that allows to capture personal big data from smartphones' internal sensors to be used for time use measurement. It allows the collection of heterogeneous types of data, enabling new possibilities for sociological urban field studies. Sensor data such as those related to the location or the movements of the user can be used to investigate and have insights on the time diaries' answers and assess their overall quality. The key idea is that the users' answers are used to train machine-learning algorithms, allowing the system to learn from the user's habits and to generate new time diaries' answers. In turn, these new labels can be used to assess the quality of existing ones, or to fill the gaps when the user does not provide an answer. The aim of this paper is to introduce the pilot study, the i-Log system and the methodological evidence that arose during the survey.

Introduction

Traditional Time Use Surveys (TUS) are essential for sociological and economic planning but face significant challenges in the modern era. Historically, these surveys rely on paper diaries where participants manually record activities at 10-minute intervals. This methodology imposes a high cognitive burden, leading to participant fatigue and high dropout rates. Furthermore, the high cost and complexity of manual surveys mean they are often conducted only once a decade, failing to capture the rapid shifts in modern behavior, such as multitasking and digital media consumption.

To address these limitations, the i-Log system was developed to leverage ubiquitous smartphone technology. By combining active user reporting with passive sensor data collection, i-Log aims to provide a "synchronic" view of human behavior, reducing participant burden while increasing data granularity and accuracy.

The i-Log System: Dual-Mode Data Collection

The i-Log system operates through a dual-mode data collection platform designed to balance information richness with user retention.

Active Mode

In active mode, the application prompts users with hourly notifications. These "tasks" ask four fundamental questions: What are you doing? Where are you? Who are you with? and What is your mood? To minimize cognitive load, responses are mapped to a standardized ontology based on WordNet.

Passive Mode

Simultaneously, the system operates in the background, collecting data from over 20 hardware and software sensors. This includes accelerometers, GPS, Wi-Fi connection states, battery levels, and system states like "Doze Mode."

i-Log notifications on the lock screen showing active tasks and background tracking.

Technically, the system utilizes a cloud-based backend with a Cassandra database for linear scalability and Apache Spark for distributed data processing. A "one-click-deployment" pattern allows researchers to rapidly deploy the infrastructure for new pilot studies.

Methodological Insights and Data Duality

A core innovation of i-Log is the "duality of data." Active diary entries serve as "ground truth" labels for the passive sensor streams. This relationship allows machine learning models to learn individual habits and eventually "fill the gaps" when a user fails to respond to a prompt.

During initial testing, researchers identified a "burden sweet spot." While higher frequency prompting (e.g., twice per hour) provided more data, it led to unsustainable dropout rates. An hourly intervention was found to be the optimal balance. Furthermore, hardware states can serve as effective proxies for behavior; for instance, monitoring a phone’s "Doze Mode" and rest states can predict sleep patterns with high accuracy without requiring manual user input.

Results from the 2019 European Big Data Hackathon

The system was stress-tested during the 2019 European Big Data Hackathon, involving 17 national teams. The i-Log dataset provided to teams included data from 66 participants, contributing approximately 190,000 hours of sensor data and 8,548 validated diary entries over a 14-day period.

i-Log application interface and system notification indicating data collection status.

Key outcomes included:

  • Integration: 6 out of 17 teams successfully integrated i-Log data into their analysis prototypes.
  • Predictive Modeling: Teams successfully built models to predict transportation modes (e.g., walking vs. bus) and identify "zones of mobility" (home vs. work) using geo-positioning data.
  • Scalability: The system handled an average of 500MB of uncompressed data per smartphone per day, demonstrating its readiness for larger national studies.
  • Innovation: The winning team from Statistics Poland developed an open-source dashboard for population time-use analysis based on these "smart" data streams.

Limitations and Future Work

Despite its success, several challenges remain. The current implementation is Android-only, which introduces potential sampling bias; an iOS version is under development. Battery life remains a constraint for older devices due to continuous background sensing. Additionally, some machine learning approaches, such as CNNs for physical activity, require larger sample sizes than those provided by the pilot study.

Future developments will focus on integrating data from wearable devices via Bluetooth. This will allow the system to capture physical activity even when the user is not carrying their smartphone, further enhancing the completeness of the time-use record.

Summary updated

Abstract

This article assesses the experience with i-Log at the European Big Data Hackathon 2019, a satellite event of the New Techniques and Technologies for Statistics (NTTS) conference, organised by Eurostat. i-Log is a system that allows to capture personal big data from smartphones' internal sensors to be used for time use measurement. It allows the collection of heterogeneous types of data, enabling new possibilities for sociological urban field studies. Sensor data such as those related to the location or the movements of the user can be used to investigate and have insights on the time diaries' answers and assess their overall quality. The key idea is that the users' answers are used to train machine-learning algorithms, allowing the system to learn from the user's habits and to generate new time diaries' answers. In turn, these new labels can be used to assess the quality of existing ones, or to fill the gaps when the user does not provide an answer. The aim of this paper is to introduce the pilot study, the i-Log system and the methodological evidence that arose during the survey.
BibTeX
@inproceedings{zeni2021improving,
  author = {Mattia Zeni and Ivano Bison and Britta Gauckler and Fernando Reis and Fausto Giunchiglia},
  title = {{Improving time use measurement with personal big data collection - the experience of the European Big Data Hackathon 2019}},
  booktitle = {RePEc:},
  year = {2021},
  doi = {10.2478/jos-2021-0015},
}
Figure