SQL analyses
Last updated
Was this helpful?
Last updated
Was this helpful?
CARTO Builder offers a UI that helps building SQL queries to perform geospatial analysis operations.
To get started with SQL Analysis, add a source to your Builder map and click on the three dots to find the Add SQL Analysis option:
After that, you will see a list of analyses compatible with your source. Compatibility depends on some factors, like whether or not the CARTO Analytics Toolbox is needed, or the connection’s cloud data warehouse.
Check this table to find out which analyses are available for each data warehouse:
Intersect and Aggregate
✅
✅
✅
✅
✅
Create buffers
✅
✅
✅
✅ (*)
✅
Add column from second source
✅
✅
✅
✅
✅
Filter by column value
✅
✅
✅
✅
✅
Calculate Centroids
✅
✅
✅
✅
✅
Clustering K-Means
✅ (*)
✅
✅ (*)
✅ (*)
✅
Trade Areas
✅ (*)
✅
✅ (*)
✅ (*)
(*) Requires the CARTO Analytics Toolbox to be installed
Each analysis will create a SQL query that performs the geospatial operation. These SQL queries will use CTEs (Common Table Expressions) to be able to chain different analyses and create a more complex sequence.
The resulting SQL queries from each analysis will take into account the syntax, specific functions and other nuances between different cloud data warehouses.
The resulting SQL query can be loaded in the map in different ways:
Run SQL analysis will load the query as a SQL Query source in Builder immediately.
Preview SQL analysis query will load the query in the SQL panel in Builder, so you will be able to review and modify the query before running it.
Save results in a new table will let you select the destination of a table that will contain the result of the query. The table will then be loaded as a source to the map.
The following are the currently available SQL analysis:
This analysis allows to perform a geospatial intersection between two different sources, aggregating data from the second source into the base one, when geometries from the second source intersect with geometries in the base source.
The result of this SQL analysis includes all columns from the base source and an extra one, called agg_value, that contains the aggregated data from the features in the second source that intersect with each row in the base source.
Parameters
Second source: pick an existing source from your map, or a table from your cloud data warehouse to be used as the second source for this analysis
Aggregation operation: select the operation to aggregate the data from the second source.
Aggregation column: select the column from the second source that will be aggregated.
Example
A very common use case for this SQL Analysis would be “Get the average revenue from all stores in each neighborhood”. In this hypothetical scenario:
a table containing the polygon geometries for each neighborhood would be the base source.
a table containing the point geometries of each store would be the second source.
the aggregation operation would be the average (AVG).
the aggregation column would be the one that contains the revenue in the stores table.
This analysis allows creating a LEFT JOIN SQL query, allowing to include columns from both base and second source to be included in the result.
Parameters
Key columns: For each source, a key column needs to be selected. This column will be used to join the rows from the base and second source that share the same value.
Columns to be included in the result: Select the columns from each source that will be included in the result.
This analysis allows to keep or discard rows based on a column value.
Parameters
Target column: Select the column that we’ll be used for the filter.
Filter operator: Select a type of filter from the list of available operators.
Values: Use the selector to configure your filter.
This analysis creates a distance buffer around your existing geometries. It works with points, lines and polygons, and the resulting geometry will always be polygons.
Parameters
Distance: Select the distance that will be used to create the buffers.
Tracts: Select the number of concentric buffers that will be created.
Individual/Combined result: Select between having an individual buffer created for each row, or combine them all in a single polygon.
This analysis will produce a point that represent the centroid of the geometries in your source. By default, it will produce a single point. Using the Categorize optional parameter we can get a centroid per category in the dataset.
Parameters
Categorize: Select a column that contains categories to create one centroid per category in your dataset.
Aggregation: Aggregate data from the original dataset into the resulting centroids. The result of the analysis will include a column aggregated_value that contains the value of the aggregation.
Aggregation Operation: Select an aggregation operation from the list.
Aggregation Column: Select a column to be aggregated.
Parameters
Number of clusters: Define the number of clusters that will be produced by the analysis.
This analysis can be performed safely with up to ~700K rows. Bigger sources can cause the resulting SQL query to hit some limits BigQuery. Due to a Redshift internal limitation, the query produced by this analysis includes a LIMIT 100 clause. Removing this limit might cause an error on the query execution.
The input source for this analysis should contain point geometries that will be taken as the origin point for the isoline generation.
This SQL Analysis is available for BigQuery, Snowflake and Redshift connections, and it requires a specific minimum version of the CARTO Analytics Toolbox Advanced module to be installed:
Snowflake: 2022.06.09
Redshift: 2022.06.07
The result from this analysis can only be saved as a new table.
Parameters
Mode: Define the transportation mode that will be used for the isoline computation.
Range Type: Define the type of range that will be be used for the isoline computation:
Distance: The resulting isoline will describe the area that can be covered by traveling a specific distance set in meters.
Time: The resulting isoline will describe the area that can be covered by traveling during a specific time set in seconds.
This analysis uses the ST_CLUSTERKMEANSfunction from the CARTO Analytics Toolbox for BigQuery, or the ST_ClusterKMeans() in PostGIS, taking a set of points and finding a defined number of clusters based on the k-means algorithm. It generates a cluster_no column that indicates the cluster that each point belongs to.
This analysis leverages the CREATE_ISOLINES function in the CARTO Analytics Toolbox for , and to generate time or distance isolines based on different modes of transportation.
