How To: Structure Data for Honeycomb ​

This guide provides an overview of the types of data Honeycomb supports and how to think about structuring your location data for Honeycomb. Watch the video or follow the step-by-step instructions below.

Supported Data Types ​

• Google Sheets*
• CSV*
• Parquet*
• GeoJSON
• KML/KMZ

*Table data format, requries certain structure

Table Data Requirements ​

Table data formats (Google Drive, CSV, Parquet) are very common. However, they are not specifically geospatial file formats, so location data within them needs to be structured in a certain way so that Honeycomb can use it.

Point Data ​

1. Basic Requirements:
   • Two separate columns: latitude and longitude
   • Must use WGS84 coordinate system
   • Automatic detection by Honeycomb
2. Limitations:
   • Supports 5-10 million points (depends on end user computer)

H3 Indexes ​

1. Format Options:
   • String in hexadecimal format (8a283082a677fff)
   • Integer representation (622203769624952831)
2. Benefits:
   • Efficient for large datasets
   • Better performance with massive data
3. Limitations:
   • Pre-aggregation required
   • Less granular than point data
   • Limited filtering capabilities

Step-by-Step Guide ​

Prerequisites ​

• Access to Snowflake
• Dataset containing location information

1. Preparing Point Data ​

1. Ensure your data has separate latitude and longitude columns
2. Convert coordinates if needed:
   • Use Snowflake transformation functions like ST_TRANSFORM()
   • Convert to WGS84 coordinate system (EPSG:4326)
3. Verify data format:
   • Correct coordinate ranges
   • Proper decimal precision

2. Converting to H3 (For Very Large Datasets) ​

1. Use Snowflake's H3 functions:

SELECT 
    H3_LATLNG_TO_CELL(latitude, longitude, 12) as h3_index,
    SUM(sales) as total_sales
FROM fact_orders
GROUP BY h3_index

SELECT 
    H3_LATLNG_TO_CELL(latitude, longitude, 12) as h3_index,
    SUM(sales) as total_sales
FROM fact_orders
GROUP BY h3_index

2. Choose an appropriate H3 resolution level for the type of data you have
3. Determine which metrics should be pre-aggregated

3. Loading Data in Honeycomb ​

1. Point Data:
   • Select table with lat/long columns
   • Honeycomb automatically detects coordinates & creates a point layer
2. H3 Data:
   • Select table with H3 indexes
   • Honeycomb automatically detects H3 columns & creates a Honeycomb layer.

Best Practices ​

Choosing Data Structure ​

1. Use Point Data when:
   • Dataset is under 10 million points
   • Detailed filtering needed
   • Individual point analysis required
2. Use H3 when:
   • Dataset is very large
   • Aggregated view is sufficient
   • Performance is priority

Tips ​

• Start with point data for smaller datasets
• Pre-aggregate to H3 for billions of records

Next Steps ​

After structuring your data, you can:

• Create visualization layers
• Add interactive components
• Implement filters
• Share maps with stakeholders

Additional Resources ​

For more detailed information, visit:

• Honeycomb Blog: Why H3 is the Missing Piece for Integrating Location Data into BI Systems
• Honeycomb Docs: Analyzing Massive Amounts of Location Data with Honeycomb
• Snowflake H3_LATLNG_TO_CELL() Function
• H3 Documentation