This project page is largely redundant and needs to be cleaned up. The main project pages for this project are:
[[Retrieving US VC Data From SDC]], and [[VC Database Rebuild]].
==Project Objective==
--113267
\COPY masdistinctkeys1 TO 'masdistinctkeys1.txt' WITH DELIMITER AS E'\t' HEADER NULL AS '' CSV
Now when you re-run the matcher on portcokeys and masdistinctkeys1 you will get 428 errors and excluding Undisclosed fields gets you down to 356.
In Excel create a column to exclude any entries where the announced date occurs before the firstinvdate. Afterwards, manually flag the other duplicate mismatches picking the earlier announced date.
==Import Flagged Matcher Data Into PSQL==
After everything in the matcher output files is manually flagged create two new tables in Excel for iposmatcherouput and masmatcheroutput and copy those files into the tables.
CREATE TABLE masmatcheroutputwithexcludeflag(
file1coname varchar(255),
file2targetname varchar(255),
warningmsg varchar(255),
file1coname1 varchar(255),
file1statecode varchar(2),
file1datefirstinv date, --mm-dd-yyyy
file2targetname2 varchar(255),
file2statecode varchar(2),
file2announceddate date,
exclude boolean,
excludemaster boolean,
excludemanual boolean,
excludemanualmaster boolean,
excludefinal boolean
);
\COPY masmatcheroutputwithexcludeflag FROM 'masmatcheroutput-nohead.txt' WITH DELIMITER AS E'\t' HEADER NULL AS '' CSV
--9400
CREATE TABLE ipomatcheroutputwithexcludeflag(
file1coname varchar(255),
file2issuer varchar(255),
warningmsg varchar(255),
file1coname1 varchar(255),
file1statecode varchar(2),
file1datefirstinv date, --mm-dd-yyyy
file2issuer1 varchar(255),
file2issuedate date,
file2statecode varchar(2),
excludedate boolean,
excludemaster boolean,
excludemanual boolean,
excludemanualmaster boolean,
excludefinal boolean
);
\COPY ipomatcheroutputwithexcludeflag FROM 'iposmatcheroutput-nohead.txt' WITH DELIMITER AS E'\t' HEADER NULL AS '' CSV
--2591
==Joining portco with ipo and mas==
In order to match the company with the ipo or acquisition date we need to join the three tables together. Do this in a two step process. First join with ipo and then take the resulting table and join with mas. This will give you companies with an ipo column and acquisition column. There will be a lot of blanks because some companies didn't have ipos or acquisitions.
FROM portcoipojoin AS portco LEFT JOIN masmatcheroutputwithexcludeflag AS matcher ON portco.coname = matcher.file1coname
AND portco.statecode = matcher.file1statecode AND portco.datefirstinv = matcher.file1datefirstinv AND matcher.excludefinal
= FALSE;
--44774
\COPY portcoipojoinmajoin TO 'portcoipojoinmajoin.txt' WITH DELIMITER AS E'\t' HEADER NULL AS '' CSV
For mas you will need to join to the masnodupes table because you will want acquiror name and transaction amount. The matcher file only contains the key without the extra information.
DROP TABLE portcomajoinmasnodupes;
CREATE TABLE portcomajoinmasnodupes AS
SELECT portcomajoin.*, masnodupes.*
FROM portcomajoin LEFT JOIN masnodupes ON masnodupes.targetname = portcomajoin.file2targetname AND
masnodupes.targetstatecode = portcomajoin.statecode AND masnodupes.announceddate = portcomajoin.file2announceddate;
--44755
Then join with the portco table again and extract out the city and address.
DROP TABLE portcoipomalocation;
CREATE TABLE portcoipomalocation AS
SELECT portcojoin.*, city, addr1, addr2, indclass, indsubgroup3, indminor
FROM portcoipojoinmajoin AS portcojoin LEFT JOIN companybase ON portcojoin.coname = companybase.coname AND
portcojoin.statecode = companybase.statecode;
--44878
\COPY portcoipomalocation TO 'portcoipomalocation.txt' WITH DELIMITER AS E'\t' HEADER NULL AS '' CSV
--now join the two ma and ipo tables together on coname, statecode, datefirstinv
FROM portcomajoinmasnodupes AS ma JOIN portcoipojoinamt AS ipo ON ma.coname = ipo.coname AND ma.statecode = ipo.statecode AND ma.datefirstinv = ipo.datefirstinv;
--44755
\COPY portcoipoma TO 'portcoipoma.txt' WITH DELIMITER AS E'\t' HEADER NULL AS '' CSV
DROP TABLE portcoipomanodupes;
CREATE TABLE portcoipomanodupes AS
SELECT *
FROM portcoipoma WHERE transactionamt IS NULL OR proceedsamt IS NULL;
--44706
\COPY portcoipomanodupes TO 'portcoipomanodupes.txt' WITH DELIMITER AS E'\t' HEADER NULL AS '' CSV
==Joining geo with portcoipomas==
Check the portcoipomas table first. Undisclosed Companies may have slipped through so you might have to rerun the joins against the companybase1 table selecting the appropriate flags. You will match portcoipomas with the geo data using coname, city, year so you will need to strip the year from the datefirstinv. Use the EXTRACT(YEAR FROM datefirstinv) command. I broke these into separate tables:
CREATE TABLE portcoipomalocationexclude AS
SELECT *, EXTRACT(YEAR FROM datefirstinv) AS datefirstyear
FROM portcoipomalocation AS portco
WHERE NOT EXISTS(SELECT * FROM excludeipomadupes as exclude WHERE (portco.coname = exclude.coname AND portco.statecode =
exclude.statecode AND portco.datefirstinv = exclude.datefirstinv AND exclude.excludeflag = 1));
FROM portcoipomalocationexclude as portco LEFT JOIN geo ON (portco.coname = geo.coname AND portco.city = geo.city AND
portco.datefirstyear = geo.startyear);
You will notice that the output has blank entries for the noaddress flag. This indicates that the tables aren't joining correctly. Indeed you will see that the company Capella Systems, Inc. in the geo table is named Capella Systems Inc. in the portcoipomalocationexclude table. Therefore you will need to use the Matcher tool to match the keys in the two tables.
CREATE TABLE geomatcher (
geoconame varchar(255),
geocity varchar(100),
geostartyear varchar(4),
file2coname varchar(255),
file2city varchar(100),
file2datefirstyear varchar(4)
);
\COPY geomatcher FROM 'matcheroutputgeo.txt' WITH DELIMITER AS E'\t' HEADER NULL AS '' CSV
--43770
Geolookupbatch1 contains the first 2299 records. Geolookupbatch2 contains 6597 records. Geolookupbatch3 contains 2527 records. This sums to 11423 which matches the number of initial lookups?