The Paper Crane Geospatial Intelligence Platform
Purpose-built for Geospatial Data Exploration and AI-enabled for Insight Discovery
Paper Crane Geospatial Intelligence Platform
Powered by GeoDataCubes™
Paper Crane is a cloud-based data platform that unifies geospatial data, generates variables through spatial feature engineering, and automates AI analysis to help you improve your models, uncover new insights and make better decisions.
The Paper Crane Product Suite
Designed to fit into your workflow, our platform is comprised of four high-performance products:
Why Paper Crane
Empowering You to Make Better Data-Backed Decisions
Modern data science has the potential to drastically improve the way we make decisions in nearly every industry and government function. However, early adopters of data-science approaches are faced with the reality that >95% of specialists’ time is spent doing critical data wrangling, feature engineering, and data-preparation before any AI analysis even occurs…. and this is even more pronounced when dealing with spatial data.
Paper Crane’s software drastically simplifies and automates this process into a fraction of the amount of time typically required. The Paper Crane user is finally free to focus on exploring ideas through the rapid iteration of model predictions, rather than wasting time and effort preparing data, battling technology, and waiting for long compute times.
What Paper Crane Does For You
Better, Faster Decisions
Better risk-adjusted policy and pricing decisions.
Amalgamate messy location and geospatial datasets in 1% of the data scientist’s time.
Lower Operating Costs
Enables data scientists to focus on modeling enhancements and reduces a company’s operating costs.
New models, products and services.
Automated lossless fusion of disparate data
- Seamlessly unify different data formats, types (vector, raster), projections, and spatial resolutions
- Automatic detection of implicit spatial data (zip codes, addresses, state abbreviations,…) and conversion to explicit spatial data via geocoding and/or linking to known spatial features.
- Lossless aggregation of attributes using location as the common key
- Integrate geospatial and non-spatial data
Paper Crane in Action
Tom is seeking strategic insights into consumer behavior for a major brand. He has several data sets including spreadsheets with sales data from existing stores, spending data linked to home addresses, regional imagery of population centers, json files with market research data at the zip-code level, and shapefiles containing polygons of high population growth.
These data are bulk ingested into Paper Crane which automatically geocodes address-level data and unifies these data with the zip-code, arbitrary polygon, and imagery data. These data are then merged with Paper Crane layers for school zones and census block group demographic data.
BENEFIT: Data wrangling is reduced to zero. Immediately import, unify, and form a single source of truth for further analysis.
Make any data geospatial-aware using CraneAI
- Exponentially increase the information content of any location data with automated spatial feature engineering
- Extend any type of data with new geospatial linkage, data appends, and extensions
- Create new custom variables using Paper Crane’s intelligent custom Feature Engineering toolkit
Paper Crane in Action
Carolyn is analyzing household exposure to various risks. In Paper Crane, she first unifies proprietary tabular data at the household level, crime data at the zipcode-level, point locations of known environmental perils, and several Paper Crane layers such as slope, elevation, climate, parcel boundaries, and flood zones.
Paper Crane’s automated spatial feature engineering module immediately computes the spatial relationships (adjacency, proximity to nearest, inclusion, density) between all entities adding 500+ new explicit variables. Carolyn then builds additional custom variables using the proximity to nearest highway, maximum slope of property, household income, the number of losses within a 5 km radius, and tree cover percent.
BENEFIT: Spatial feature engineering is a crucial element of a data science, but it can be a specialized and time-consuming task. Paper Crane does this nearly instantaneously.
Customize and refactor spatial data for optimal utility
- Make spatial data tabular and AI-ready at any chosen feature resolution using GeoDataCubes
- Extract unified data in any format at any spatial feature aggregation level with zero data loss.
- Automated imagery/raster incorporation and transformation in real-time
Paper Crane in Action
José is evaluating different commercial real estate sales prediction models. These models require similar inputs, but one model operates at the parcel level resolution, one at the census tract level, and one at the county level. He also has a spreadsheet of addresses with past sale prices for selected properties.
Using Paper Crane, Jose merges his data with the Paper Crane layers necessary for the data-hungry models. With one click, he exports these as tabular data at the property-level, the census-tract level, and the county level for input into his models. All zonal statistics, spatial feature engineering and aggregates of the underlying variables are automatically handled by Paper Crane making cross model analysis simple.
BENEFIT: The lossless geodatacube structure enables the effortless transformation of data into the spatial resolution or feature level appropriate to the task at hand.
Analyze and Model
Explore data relationships and predict outcomes using CraneAnalytics or the modeling and visualization platform of your choice
- Discover hidden data relationships and trends
- Find new predictors with automated variable importance analysis
- Leverage automated machine learning and deep learning pipelines designed specifically for geospatial to build deployable models
Paper Crane in Action
Claudia’s team is designing a new property-level wildfire insurance product. Paper Crane ingests her proprietary insurance claims data and a set of post-event building impact data as training data. She then appends ~800 paper Crane building-level, property-level, proximity and community variables to each address in her training set.
She then either imports her prepared tabular data directly into her preferred AI analysis package or optionally keeps these data within Paper Crane product ecosystem to take advantage of the Paper Crane automated predictive modelling suite. Within the modelling environment, top predictors of fire damage are identified, and a genetic algorithm automatically constructs the optimal AI machine learning model for wildfire damage.
BENEFIT: Gain a critical edge in predictive modeling by using AI software purpose-built for huge data volumes, large numbers of variables, spatial awareness, and rapid iteration.
Making Geospatial Data AI Ready
- Patent-pending partitioning and recursive subdivision algorithm transforms spatial data into this mathematical representation.
- Enables expensive spatial calculations to be performed as real-time linear aggregates.
- Enables real-time spatial data exploration, analysis, automated machine learning, and automated deep learning on heterogene
Converting pixels and polygons to actionable insights in minutes.
- Unparalleled combination of patent-pending GeoDataCube architecture, extreme degree of automation, and specialized AI (machine learning and deep learning) pipelines tuned to the geospatial domain.
- Leveraged throughout the Paper Crane Platform to enhance data and provide insights.
- Applicable in insurance, real estate, agriculture, government, retail, digital marketing, and many other industries.
- User-accessible GUI available Q4 21
“Your work has been beyond stellar, and you never cease to amaze. Talk about over-delivering, these are great additional features beyond what we discussed! Still blowing expectations out of the water!”
– Data Scientist at a Top 15 Insurance Company