SELECT FROM cosvec-target
Before you begin
- SELECT examples
- SELECT statement
- COSINE_DISTANCE and EUCLIDEAN_DISTANCE functions
- CREATE TABLE cosvec-target
- INSERT INTO cosvec-target
Comparing distance metrics
FeatureBase supplies two functions to measure VECTOR data type values:
- cosine_distance measures the similarity of angles between two vectors, and is unaffected by their magnitude
- euclidean distance measures the distance between points, rather than the angle between vectors
The following example illustrates the difference between these measures:
SELECT cosine_distance([1.0, 0.0], [0.0, 1.0]) as cos_1_0,
euclidean_distance([1.0, 0.0], [0.0, 1.0]) as euc_1_0,
cosine_distance([2.0, 0.0], [0.0, 2.0]) as cos_2_0,
euclidean_distance([2.0, 0.0], [0.0, 2.0]) as euc_2_0;
cos_1_0 | euc_1_0 | cos_2_0 | euc_2_0
---------+-----------+---------+----------
1 | 1.4142135 | 1 | 2.828427