data_validation_smaller_table.sql

create or replace table "SUPPLIER_ORIGINAL" as
select * from "SNOWFLAKE_SAMPLE_DATA"."TPCH_SF1"."SUPPLIER"; -- 10000 suppliers
create or replace table "SUPPLIER_FEW_CHANGES" clone "SUPPLIER_ORIGINAL";

-- introduce some errors
select count(*) from SUPPLIER_ORIGINAL
where s_address like '%Z%'; --3190 rows will be modified

update "SUPPLIER_FEW_CHANGES"
set S_ADDRESS = replace(S_ADDRESS, 'Z','z');

select approximate_similarity (mh) from
  (
    (select  minhash(100, *) as mh from "SUPPLIER_ORIGINAL" )
    union all
    (select  minhash(100, *) as mh from  "SUPPLIER_FEW_CHANGES" )
  ); --0.55