For Researchers
Location data looks like the easy part of a time-use or travel study: put a logger on a phone and you have everyone’s day. In practice, a GPS trace is thousands of noisy points with gaps in them, and turning it into something you can analyze is a research project of its own. Daynamica does that work as the data comes in, and asks participants about the parts no sensor can see.
Illustration with made-up data; no real participant’s movements are shown.
Each of these sits between raw location data and the first table in your paper. With a GPS logger, they’re your team’s problems. With Daynamica, most are handled before the data reaches you.
Raw GPS: Signals bounce off tall buildings and drift while a phone sits still, so a person at their desk appears to wander around the block.
You’d need to: filter outliers and smooth the trace, choosing thresholds that decide which movement was real.
Daynamica: detects activities and trips from GPS and the phone’s accelerometer together, so detection doesn’t rest on position alone, and participants confirm each episode.
Raw GPS: Signal drops inside buildings and underground, phones cut back location updates to save battery, and people leave phones on a desk or a kitchen counter.
You’d need to: decide what happened during each gap, or throw the day out.
Daynamica: participants see their day as a sequence of episodes and confirm or correct it, and an end-of-day review can ask about what sensing misses, such as time at home or work while the phone wasn’t on them.
Raw GPS: A cluster of points is just a cluster. Five minutes in one spot could be a red light, a bus stop, or a coffee.
You’d need to: pick a minimum dwell time and distance, knowing that the choice changes how many trips and activities you count.
Daynamica: divides each day into the alternating activities and trips that time-use research expects, and shows them to the participant to confirm or correct.
Raw GPS: Speed alone confuses a bus in traffic with a car, and a walk to the bus stop with the bus ride itself.
You’d need to: build or borrow a mode-detection model, then split trips at every transfer.
Daynamica: tells most major modes apart from sensor data alone (walk, bike, car, and more), and the participant confirms the exact mode with one tap.
Raw GPS: Coordinates can’t tell a clinic visit from a pharmacy run, a friend’s house, or the café next door.
You’d need to: match locations against map data and hope the match is right, or survey participants separately and line the answers up afterwards.
Daynamica: participants classify each activity from a short list set for your study. Once they tag a place, later visits there are classified automatically, and after 5 to 7 days of tagging, days can typically be reconstructed with more than 90% accuracy from passive sensing alone (Fan et al., 2015).
Raw GPS: Every inference above is a guess, and the trace carries no record of what the person was doing or how it felt.
You’d need to: run a separate diary or recall survey, with its own burden and recall bias, and reconcile the two.
Daynamica: asks short questions tied to specific episodes, such as the type of care and the wait at a medical visit, while it’s fresh. Participants’ confirmations are the ground truth.
Daynamica isn’t just better GPS processing. It’s built on the recognition that sensors and people are each good at different parts of the job, and it combines three layers:
Because repeat visits to tagged places are classified automatically, participants do most of their tagging in the first week. That’s what makes month-long studies practical: the burden falls as the data accumulates. You can see the layers in action on How It Works.
| Step | GPS logger | Daynamica |
|---|---|---|
| Cleaning noisy points | Your team | Handled as data is collected |
| Filling or explaining gaps | Your team, by assumption | Confirmed by participants; end-of-day review |
| Finding stops and trips | Your team, with dwell-time thresholds | Detected automatically, confirmed by participants |
| Travel mode | Your team, with a mode model | Detected automatically, confirmed by participants |
| Activity type and purpose | Map matching or a separate survey | Tagged by participants; repeat places automatic |
| Experience and context | Not captured | Episode-level and end-of-day surveys |
| What you analyze | Points you’ve processed yourself | Episodes, with surveys attached, ready for analysis |
The Daynamica Data Explorer runs on an anonymized sample of Daynamica data: the episodes and survey answers your team would receive, not raw points. Working With Daynamica Data explains how the export reaches you and the open-source tools for analyzing it.
Tell us what you’re trying to measure, and we’ll show you what the data would look like for your study.
DAYNAMICA