PanTera works best when you can explore the data yourself, so we built an export function that allows PanTera users to create fully interactive representations of their data in an easy to distribute and share embeddable web format.
For a while now we have been showing the 2013 New York city Taxi data set, because it has rich geographical and financial data in a context that is easy to understand. Seeing the data in pictures and videos is one thing. Being able to explore it is another.
Today we are presenting the first interactive, embeddable visualization made with PanTera showing pick up and drop off locations of all 187 Million taxi trips made in NYC in 2013.
Some of the interesting features we found in the data are:
- Blur in and among the tall buildings contrasted with great street alignment in the burroughs dues to GPS signal reflection in the urban canyons
- High-level (city wide) location distribution patterns:
- Street level patterns, drop offs are relatively uniformly distributed while pickups in the boroughs are heavily skewed towards routes (and even the side of the street) headed into Manhattan
- Other findings:
Explore the data, have fun and let us know if you find something interesting that we missed!
We will be using this space to showcase some of the interesting things we discover as we work on PanTera. This may include new technical approaches or interesting findings as we explore datasets. To begin, we just want to provide some background on why we’re doing what we’re doing.
We created PanTera because existing visualization tools are not tailored to meet the specific opportunity and richness presented by large amounts of data. Conventional approaches to the analysis and visualization of big data involve running queries, sampling, or averaging information. These techniques lead to a loss of nuance and may obscure patterns that don’t fit neatly into simple bar and line graphs.
The following principles guide our work on PanTera:
Great Visualization Matters
- We are committed to harnessing the power of human perception
- Innate human capabilities can be used to perceive and process data rapidly and efficiently
- Visualization translates data into a visible form, highlighting important features, including commonalities and anomalies
- Not all visualizations are created equal
- Human performance can vary significantly - plus/minus 100 times according to the representation used 
- Information visualization techniques amplify cognition by 
- Decreasing the load on mental resources 
- Reducing search times
- Improving recognition of patterns
- Increasing inference making
- Increasing monitoring scope
Big Data, Small Nuances
Data visualizations have not evolved to meet the needs of big data analytics. Traditional methods of using SQL and charts from the 90s have been applied to big data by funnelling the data down into simplified plots. The current methods of sampling or averaging data lead to scaled-down representations of information.
Given the amount of time and money used to collect and store data, we believe that a tool that allows seamless engagement with the entirety of datasets is critical. By preserving detail from a global scale down to individual data points, PanTera allows one to derive unique insights from big data.
Innovation is Key
We decided to not take the traditional path using SQL queries to create summarized plots of data. We are changing the way one approaches data so that it is possible to plot it in its entirety, thus ensuring as many insights as possible are revealed.
As we continue down this path to meet the visualization needs of big data, we are excited to see all the beautiful things data has to offer.
Did we mention PanTera makes data beautiful?