Datamining and Point of Diminishing Returns?

[aflaba]

No real benefit.

It's just that as someone said they have a maximum size of 2 GB.

How it is accomplished:
First I import observed hands into one database.
When it is full I create a new database and start importing into that one.
Not tricky at all.

[Professor Frink]

hey go to my comp -> c: -> program files -> poker tracker folder -> click (or right click) on the ptrack.mdb file and it should show the size

for an indication my database is 90,000 hands and is 502Mb whilst the hand history database (hhdb.mdb) is 288Mb

hope that helps

[nevadaJACK]

[ QUOTE ]
hey go to my comp -> c: -> program files -> poker tracker folder -> click (or right click) on the ptrack.mdb file and it should show the size

for an indication my database is 90,000 hands and is 502Mb whilst the hand history database (hhdb.mdb) is 288Mb

hope that helps

[/ QUOTE ]

I see....but just what is the benefit of even hanging onto one's hand-history database...I usually just delete my HHs from my computer after they have been added to my PT database??

[mosta]

[ QUOTE ]
Do most of you use completely separate databases for your DM efforts?

Freakin

[/ QUOTE ]

I keep hands that I played in a separate database from where I keep observed hands. That way the db where I review my numbers doesn't become clogged and bogged down with irrelevant hands slowing it down. it used to also be important to have your hands separate for purposes of calculating rakeback, but they've worked around that problem. if you keep everything in one db, and you fill it up in a month or two, then you can't review your play over several months, or longer, in one ring game stats window. I also keep separate db's by limit and by short or full table.

one disadvantage of multiple db's is that when you run your advanced export of statistics you can not use the new feature to use only the most recent N number of hands per player. it's too complicated to figure the most recent up to that number if it has to compare across multiple databases. you can limit the hands it uses by hands only since a certain date, but that is not as useful a restriction. the idea is that you want enough hands to get accurate numbers, but you also want to see only their current playing style in case it's been evolving, so you don't want a bulk of older hands to average out new playing characteristics. the best way to do it is to look at the last 300 to 500 or 1,000 hands only. cutting it off by date is not as effective, because that throws out hands you want from players you haven't seen in a long time, and keeps more hands than you need from players that play a lot.

http://www.pokertracker.com/forum/vi...amp;highlight=

[Rudbaeck]

[ QUOTE ]
2GB is the limit for the pokertracker database. Pat is working on a new version using MySQL which will not have this limit. However, I find that after about 1GB, performance suffers.

[/ QUOTE ]

Large Access dbs are horrible. An equal sized MySQL db will be alot faster.

But there is light at the end of the tunnel.

1) Buy a 15krpm Ultra320 80-pins SCSI drive. You can find a 40gb for ~$200 or so. With an average access time of 3.9ms it's nearly twice as fast as the fastest IDE drives. It can also load about 960mbit/s.

2) Install PT on this drive, and never put anything else on it. Defrag often, compact database religiously.

Even a spare regular IDE drive solely used for PT is better than it sharing drive with the poker clients. You really don't want to read from the same drive you are writing to.

With a 1.5gb database you should still be able to import 8-12 tables worth of hand histories every minute without it using up more than a few seconds with a setup like this.

[Rah]

You are talking about diminishing marginal returns. Do you understand what the phrase "diminishing returns" means?