Scaling Visual Analytics with HEAVY.AI

Scaling Visual Analytics with HEAVY.AI

‍

In today’s data driven enterprise economy, the impact of interactive visual analytics can differentiate an organization’s data offerings. However, persistent issues with secure and efficient data access complicate matters. Conventional tools are slow and unsuitable for the digital era and have limited use and value. To address these limitations, data producers can create their own delivery platforms. This involves developing and maintaining a custom software stack, requiring significant investments in software engineering and user experience design expertise, especially as traditional back-ends lack real-time interaction capabilities with large datasets. DIY solutions often have limited architectures, hampering scalability for user volume and data magnitude, especially for conventional analytics and geographic information system (GIS) solutions relying on data movement for comprehensive charts or maps.

‍

Solution: Scalable Back-End Rendering

‍

HEAVY.AI's robust platform offers scalable back-end rendering, creating visualizations on the server rather than transferring extensive data to client devices. This benefits data producers and consumers by delaying data movement until user-driven subsetting is complete, ensuring visually appealing and secure customer data even on smaller devices or unstable networks. What’s more, leveraging cloud provider servers enables customers to seamlessly scale with business growth or demand fluctuations.

‍

Back-end rendering eliminates complex front-end development and data chunking. A small HEAVY.AI cloud instance easily manages map renders and charts with millions of features, while medium instances handle datasets with billions of features. Data remains secure until download, and clients receive runtime-rendered or auto-generated aggregates, enhancing the user experience. Embracing HEAVY.AI enables enterprises to enhance data offerings while navigating challenges.

‍

Four Progressions Towards Enhanced Value

‍

The evolution of online data delivery solutions generally unfolds across four distinct stages with gradual value enhancement: 

‍

1. Achieving Seamlessness

‍

Data providers aim to stand out and expedite insights, but data delivery presents challenges. Excessive file sizes create an unwieldy user experience, especially with geospatial data and spatial tiling, introducing abstraction barriers. Locating and comparing data across tiles and time frames becomes complex. Existing methods, for example, cloud object storage, struggle with big spatial or temporal data distribution, leaving users to manage that complexity. Graphical interfaces offer improvement, but their construction is intricate. Legacy technologies require arduous tile map setup and downloads, and this fragmentation impedes customers from deriving immediate value from data, highlighting a significant initial barrier.

‍

2. Unleashing Interactive Slicing and Dicing

‍

Data value and adoption suffer from producer-driven tiling or file divisions, while consumer-driven segmentation fosters increased value. Consumers typically focus on data facets, geographic subsets, or temporal ranges, and the adaptability and variation of such subdivisions enhances data utilization. Smaller datasets allow manual extraction post-download, while larger ones benefit from deferred downloads until visualization, adding substantial value. OpenSignal, a crowdsourced data provider for telecoms companies and HEAVY.AI OEM partner, exemplifies this value proposition by offering customers its Explorer platform, providing access to 10 billion daily dataset measurements, prioritizing "time to value." OpenSignal achieved this functionality through straightforward customization of HEAVY.AI's pre-built features.

‍

3. Interactive Scenario Planning

‍

Moving up the value chain, more HEAVY.AI clients are using interactive analytics for online user-driven scenario planning, accommodating business uncertainties and exploring multiple possibilities. For example, a HEAVY.AI customer that enables cities, states, and utilities clients to create scenarios to guide energy policies and meet carbon goals, generated interactive dashboards that require robust interactive performance despite numerous scenario combinations. They leveraged HEAVY.AI's GPU analytics to generate combinations on-demand, eliminating pre-computation needs, highlighting the importance of purpose-built front-end designs for supporting computational back-ends in enhancing interactive scenario planning.

‍

4. On-Demand Predictive Analytics

‍

Expanding on scenario analyses, data providers can leverage predictive analytics. Another HEAVY.AI customer, an energy data provider, employed the technology to combine historical data and user inputs for predictive analysis, offering faster, more accessible insights than traditional tools. This organization customizes HEAVY.AI's back-end analytics integrating predictive capabilities and providing interactive displays and predictive functions, a high-value solution.

‍

The customer relies on standard HEAVY.AI Immerse front-end features while adapting back-end analytics, showcasing HEAVY.AI's architecture's adaptability for various predictive analytic applications, expanding possibilities for data providers.

HEAVY.AI’s software enables customers to offer enhanced and differentiated data capabilities through interactive analytics without the limitations of DIY approaches. Replacing cumbersome download-based workflows with user-driven exploration adds value at no extra provider cost. More advanced offerings require industry-specific effort, aligning with customer value and building on a data foundation, and leveraging industry knowledge provides specific insights on the back end, while customizing the user experience fosters deep customer engagement on the front end.

‍

‍