MapD is 2.0

After many months of hard work, refinement and improvement, we’re very happy to announce the release of version 2.0 of the MapD Core database and Immerse visual analytics platform. This is the result of a valuable collaboration with our users and customers since our public launch in March, and is designed to give them even more of the analytical power, ease of use and GPU-accelerated speed that they’ve become used to in our first version.

Today we want to walk you through some of the improvements we’ve made to our Immerse visualization client, with more to come in a second post on our backend, the MapD Core database.

By running on a backend as fast as the MapD Core database, we have an opportunity to build a client in Immerse that looks deeper into datasets to extract meaning, considering data from more dimensions than would be feasible with most systems. Where most analytic systems need to marshal their resources carefully to avoid bogging down the user experience with heavy queries, in Immerse we can give the user concurrent, multi-dimensional views into their data because of the high speed available to us via the GPU. This performance has allowed us to design Immerse for maximum emphasis on instant data exploration, without the need to tiptoe around computing resources. The result: a system which is both fun to use and can truly “immerse” our users in insights from their data.

So, what are the new features?

First, Immerse continues one of our users’ most popular features, the cross-filter paradigm, which causes filtering one chart to also filter the data shown in all other charts on a dashboard. This makes the process of data exploration incredibly interactive, and we have added technical refinements in v.2 to further speed the experience, by making the client fully asynchronous. This means that data can be given back to the user as soon as it is ready, without one query blocking the another.

Second, since Immerse is about flexibility and speed in data exploration, we also wanted it to be flexible and quick to set up a dashboard. This is why we’ve eliminated a “chart-first” setup, where you needed to begin by specifying a chart with no option to subsequently change it. Instead, we’ve moved to an open chart creation experience, where you can build visualization in one chart type, and then move to any other data-compatible chart type to see which visualization you prefer. As shown below:

Third, we’ve improved the use of data-driven colors in the product, so that in addition to being able to color charts by quantitative measures, you can now choose to assign custom colors to your categorical information. For example if you want the slices in a pie chart to be red for Verizon, pink for T-mobile and yellow for Sprint, you can now make it so.

Fourth, we’ve made it easier for you to understand the layout of your numerical data by introducing quantitative binning, which gives you histogram-like functionality on most chart types. This lets you study the distribution of your data across a numerical measure by aggregating the data into bins of your chosen size.

Fifth and finally, we’ve begun the introduce the power of the MapD Core database by allowing custom SQL to be used to shape your data into units which will be most revealing for your analysis. Quick chart building is accessible via intuitive chart controls and filters, but even more control can be taken by writing SQL statements to filter and group data as you wish. An example:

While version 2.0 is a major step forward in our product, we feel that MapD is only just getting started in bringing the power of the GPU to the data analytics world. We invite you to try out the new version our product at on our demo page and let us know what you think. We’re excited to keep pushing the boundary on the revolutionary data analytics capabilities now available to us from the GPU. In our next post, we will delve more into how we’re leveraging this power in version 2.0 of the MapD Core database.