Dec 2016 · 4 Year and 11 Months

MovArt

This was a project inside my masters Degree. The goal of the proposed project was to create an artistic and visual representation from a large amount of data. So I decided to do a visualization from all the data from a person's Google Timeline.

projects/movart

The main goal of this course was to do a representation of the data, without being clear what it meant, but the representation had to come from real data. So my work was focused on two big aspects, how I was going to represent the data and understand what data I had and explore what kind of visual variations I could do.

For the visualization I did some research, to understand what kind of would be used to represent each kind of data.

Visualizations Reseach

So I came up with the data from the timeline of a Google account, that at the time, was not offered by the Google API, but each user could download his data as a JSON file.

I used several languages to implement this project. For the visualization, I used Processing, a JAVA based language that was lectured in the classes, receiving and processing the JSON file I used the languages PHP, HTML, and Javascript.

After analyzing the JSON file from Google Timeline, I was able to categorize it. This data was organized in a set time and Google predicts the kind of activity you were doing at that specific time at a specific location. These predictions varied from the activities “walking”, “in a vehicle”, “bicycle”, “tilting” and “standing still”.

So I had the list of different activities with a set time with a location and I had to show a different way of representing them.

Symbols designed to represent each data classification

But I could get more information from this list of activities. For example, if in two chronologically consecutive activities and the location was exactly the same, it meant I stayed at the same place between the two time periods of the activities. With this thought, I could get how long I stayed at each location that I decided to represent by the size of the shape, the longer the period, the bigger the shape.

On the other hand, if in two chronologically consecutive activities and the location changed, it meant that I dislocated myself and the kind of movement was defined by the second activity, “walking”, “in a vehicle”, “bicycle”.

At this point, I had a group of data lists of activities with a location, timestamp, duration, and a list of dislocations. I could start working on the visualization and processing of the data.

First I started by grouping the data by day, so I drew my daily activity. I was going to draw every activity in a container and I used each activity position to distribute the shapes across it.

I could have distances between activities of 1km in some days and other days 100km and I wanted in both cases to be an even distribution of the shapes across the container. So, I made the decision to deform the proportion between latitude and longitude to fit perfectly into each day, since the clarity and transparency of the data was not a goal for this project.

Visual representation of one day

Later in the project, I started incorporating new features, such as when you select a day, when you go to see the visualization, the shapes are shown in a timeline animation, giving a better perspective of the changes and activities across each hour of the day.

Another feature I added was to group several days, so the user could compare visualizations of different days in a week. For example, in some visualizations of my weeks, during the weekend I stayed more time at home, so I viewed less movement and bigger forms. On other days, where I had more movement than normal I could pinpoint to the days I was out with friends or playing Pokemon Go.




This was a very fulfilling project. I always take a lot of pleasure dealing and working with large quantities of data, organizing it with code and group and designing this data to get interesting conclusions from it.