The Four Mes

⚖️ Ethics & Responsibility

🔒 Privacy Statement

This dashboard contains my own location data, knowingly published by me, the data subject. It identifies my home (Punna Residence 5) and primary workplace (Choei) at building precision. I'm comfortable with this trade-off because (a) I'm both the subject and the publisher, and (b) the academic value of the project is worth the disclosure to me. Do not replicate this approach with someone else's data. The dataset contains no third-party names, no contacts, no timestamps, and no biometric data.

⚠️ Bias & Limitation Disclosure

• No timestamps. Bump strips them,no time-of-day, no day-of-week, no recency. Every pattern claim here is spatial, not temporal.
• "Log" ≠ "visit." A log is a GPS ping inside a ~65 m hex. Driving past a restaurant counts as a log. Don't read the counts as "times I went there."
• GPS noise. Phone GPS has ~5–15 m error indoors and near tall buildings,pings can land in the neighbouring hex.
• Indoor under-counting. Bump logs less reliably indoors, so cafes I sit in for hours may look like a single ping; streets I walk get more.
• Geographic scope. Only Thailand. Budapest is in the story but invisible on the map.
• Tag selection bias. 87 places are tagged out of thousands visited. I tagged what I cared to remember,survivorship bias.
• Persona self-categorisation. I assigned places to personas based on self-knowledge, not a clustering algorithm. Subjective.

📊 Visualization Justification

• 3D hex extrusion. Height encodes log count so you feel intensity. Hexagons (H3 grid) avoid the uneven cell sizes you'd get from squares projected at this latitude.
• Turbo colormap on a log scale. Linear color would crush the 3,000 low-count hexes into one dim band; log brings out the texture without overstating differences.
• Scratch trail vs. towers. Two visual languages so "I've been here" (pink flat) and "this place matters to me" (3D tower) don't get conflated.
• Dimming non-key towers per persona. Keeps the dimmed towers visible at low alpha so persona view stays geographically honest,I'm not hiding data.
• Risks of misreading. Tall column ≠ "I love this place." Could just mean it's on my route. The Delhi Street Indian Restaurant in the dataset has 78 logs and I've only eaten there once.

🎯 Responsible Decision

Based on the data, the honest decision I can make is: "My geography is 80% habit. If I want a wider life, I have to plan for the other 20%." The data doesn't tell me whether that's good (habit is also home, work, friendships), only that it's true.

What I won't claim: that I should go out more (the data has no happiness variable), that any tower is a "favorite place" (it might just be on a route), or that Adventure Me is "underdeveloped" (one weekend in the mountains could shift the count dramatically,small dataset).