Line 1: Player MLBAM BDB Retro
Line 2: correct STATS, but has incorrect in Xref_Stats

Chris Smith 434672 smithch07 smitc002
8253 not 6378

Jacque Jones 150218 jonesja04 jonej003
6246 not 7191

Tom


---------------------------------------------
The Book--Playing The Percentages In Baseball
http://www.InsideTheBook.com

In conjunction with the earlier mention,  jonesja05 (Jason Dewey Jones)
should have a STATS Id of 7191 not 6246.  Basically, the two were swapped.

Tom



> Line 1: Player MLBAM BDB Retro
> Line 2: correct STATS, but has incorrect in Xref_Stats
>
> Chris Smith 434672 smithch07 smitc002
> 8253 not 6378
>
> Jacque Jones 150218 jonesja04 jonej003
> 6246 not 7191
>
> Tom
>
>
> ---------------------------------------------
> The Book--Playing The Percentages In Baseball
> http://www.InsideTheBook.com
>
>


---------------------------------------------
The Book--Playing The Percentages In Baseball
http://www.InsideTheBook.com

I have found a willing partner in MLBAM to provide all their IDs.  I have
taken the first step of cobbling ID maps of various people and sources,
and have created the following:

http://www.insidethebook.com/ee/images/uploads/EXPORT_ID_MAP.zip

More details are here:
http://www.insidethebook.com/ee/index.php/site/comments/the_universal_player_id_\
and_biographical_data_project/

For all of you who have an ID mapping file, PLEASE, download the above,
link it to yours and:
1. fix your files if you find errors
2. tell me about any errors I have
3. after you fix your errors, submit your ID mapping file to me; I can
certify it as accurate, or it might help me in finding more errors

I have so far cobbled data from several sources.  My hope is that after 2
or 3 iterations of this that I will come up with the definitive ID mapping
file.

If someone out there works for STATS, then be a good guy and help me out.
The BIS data looks like it's complete, as is the Retro and BDB.  There is
still some problems with the MLBAM data (a few dozen corrections
required).

Once this is done, we can then link to the biographical information (name,
handedness, birth, death, school), and MLBAM as I said is a willing
partner here.  MLBAM includes not only the 17,000 or so MLB player ID, but
another 65,000 or so minors and others player IDs.

Hopefully, at some point in the near future, we can make it so that we
have overnight updates any time corrections are made.

Help me, help you.

Thanks, Tom

---------------------------------------------
The Book--Playing The Percentages In Baseball
http://www.InsideTheBook.com

I forgot to note that these are showing duplicates based on conflicting
data sources and I haven’t tried to resolve them yet:

bis_id mlbam1 mlbam2
18 297292 491703
49 112004 452035
263 110638 445216
936 233594 446248
1137 237796 425532
1479 400123 420944

stats_id mlbam1 mlbam2
6918 400123 420944

Tom

----- Original Message -----

From: Wells Oliver

To: baseball-databank@yahoogroups.com

Sent: Thursday, February 04, 2010 4:55 PM

Subject: [baseball-databank] Extra base hits for pitchers?

 

Am I missing something or is this not included in the Pitching table, nor derivable- there's no doubles nor triples.

--
Wells Oliver
wells@submute.net

Paul DuBois, "MySQL Cookbook," recipe 1.2 creates a database of limbs.  Humans
have 2 arms and 2 legs.  Insects get 6 legs and 0 arms.  A phonograph gets 1 arm
and 0 legs.  The only "thing" that gets NULLs for both arms and legs is the
"Space Alien" because of its unknown physiology.
I think that Sack hits and flys were zero rather than NULL.  This opinion holds
for many other of the categories in the data bank tables.

> Paul DuBois, "MySQL Cookbook," recipe 1.2 creates a database of limbs.
> Humans have 2 arms and 2 legs.  Insects get 6 legs and 0 arms.  A
> phonograph gets 1 arm  and 0 legs.  The only "thing" that gets NULLs for
> both arms and legs is the "Space Alien" because of its unknown physiology.
> I think that Sack hits and flys were zero rather than NULL.  This opinion
> holds for many other of the categories in the data bank tables.
>
>

Ted is correct in his other response.  We want to classify things as:
a - known to not have occurred
b - unknown to have occurred
c - impossible to have occurred

So for a), we represent this as "0".  For c), we represent this as "null".
  There is 100% agreement here.

The other item is where the issue occurs.  If you make b) "0", then how do
you distinguish it from the legitimate "0" in a)?  You can't.  If you make
b) "null", then how do you know if it could possibly exist or not?

Say, I dunno... batter-balks used to exist, and now they don't, and you
have a category called batter-balks.  You set it to null for all the
present-day players, and, what... 0 for the other guys who played in the
batter-balk era but have unknown totals?  No, you make them null.  But,
then how do you distinguish the batter-balk era from the other era?  You
can't from this particular data, but that's why you have some other table
that defines the season range for the data.

So, you can have a complete description of the data by only using 0 when
you know the total actually was 0.

Tom




---------------------------------------------
The Book--Playing The Percentages In Baseball
http://www.InsideTheBook.com

Tango Tom and other interest baseball data bank users,  Wednesday,

You really believe that the present NULL defaulting plan up is justified and
most correct. I have given up all hope of convincing you that anything else
should be considered.

Perhaps you will consider some other method of making the historical record
consistent.  To that end, I have uploaded, (to the files section of this group,)
a file named Tom.txt.  That file records the results of the same query about
SF's, (sacrifice flys,) to BOTH the present NULL defaulted database "bbdb2," AND
a zero defaulted clone of that same information, "bbdb1."

If you study the results you can see some of the early years have zero SF, (sack
Flys) and some have NULLs.  Shouldn't your strongly held argument apply
consistently to all data displayed for these years?

Notice how years since 1954 are identicle.  Indeed after studing many trends
spanning these years using both methods of defaulting, I see major advantage in
defaulting to zero.  Perhaps I have not seen a disadvantage only because of the
things I study.  So far my observation remains, defaulting to zero produces
better trend analysis.  But there I go again trying to win you over!

I know that this topic smells of Formaldehyde to most Data-bank users.  But
since the data-bank is largish and is comprehensive, it is useful for students
of certain arcane statistical processes.  My hope is to make useful analysis of
long term trends.  So I have and will make available either the code to modify
the current database, or a derived database for those who might request it.

Bueno Venturs, we have come to a parting of the ways.
Cactus Mitch,

Sean, Tom,

Yes it is easy to change NULLs to zeds.  I've got lots of practice.  Then I
decided that it would be easier to alter the defaults.  Then I noticed that in
some cases the groups selected were different.  MySQL handles NULLs and ZEDs
differently when selecting or displaying results based on computations.

I have been trying to recreate Gould's conclusion that baseball is getting
better and that the demise of the 400 hitter is a result of that improvement, I
noticed a quantity effect.  Increasing the number of at bats decreases the
probability of maintaining a high batting average.

Gould had a computer and a programmer available though he did much of
computational work in his hospital be.  He commented on baseball statistics
medicinal value.  Never the less he used a high low-end cut- off of about 100 at
bats, (if I remember correctly.)  That seemed worth study to me.  Early on there
were many fewer games and therefor fewer at bats.   In a drug effectiveness
study one should not de-select subjects as inappropriate if they show adverse
reaction tendencies and just say they were not in the experimental group.

Why is the post-season batting table Home Run column defaulted to non-NULL and
zero in the present offering?  The way I suggest is in the data base already,
albeit in only one column that I have found so far.

Some might point out that we are dealing only with historical records.  
Devotion to baseball stats is not for everyone.

Catchers and pitcher report soon. cactusmitch

Walko-san wrote:

> I asked Michael Westbay of JapanBaseball.com to host it, since I figured it's
probably the best English language hangout for Asian baseball fans:
http://japanesebaseball.com/data/index.gsp

The original data set is at the above location.  The links to the 2009
data can be found here:

http://www.japanesebaseball.com/forum/thread.gsp?forum=3&thread=30250#377608

Hope this helps.

--
Michael Westbay
Writer/System Administrator
http://www.japanesebaseball.com/

> Why is the post-season batting table Home Run column defaulted to non-NULL
> and zero in the present offering?  The way I suggest is in the data base
> already, albeit in only one column that I have found so far.

If you followed my rules, then you would have to conclude that there is
never an occasion in playoff baseball that someone hit a HR and we didn' t
know about it.  Hence, nulls are impossible.

It seems to me that taking a stance that HR is non-nullable in playoff
history is perfectly fine.

Tom

I'm still learning how to use the up load features of other parts of this
group....
On the Open Office spreadsheet it is pretty easy to change the display number of
decimal point accuracy, but that didn't seem to come all the way through to the
Yahoo database file.  The graph, (in the Photos section, (I used R,)) kept the
more accurate display easily for the study using Zedded entries.

I still think that is best to convert the NULLs to ZEDs in most cases.

Ab Bats per Plate Appearance or Batting Opportunity as some call it is
declining.  The current scoring rules advantage pitchers over hitters.  I like
watching low scoring pitching duels.  But is it "Better Baseball," and is it as
exciting for kids?  I don't even talk with my grand kids about such like.  My
grand daughter likes rugby!

There is no FAIR comparison of todays batters with those of old.  Sad but true.

I just ran a query on all the colleges that players attended.  The numbers are
lower than the ones from B-ref.

In a couple of case, the player attended two universities, but in most cases the
college is just missing.

Is there a reason for this discrepancy?

Gabe Molina is an example.

The definitive will be in this thread:

www.insidethebook.com/ee/index.php/site/comments/the_universal_player_id_and_bio\
graphical_data_project/

I'm still working on it.

Tom

> Sorry, is there a definitive place to go (or an almost definitive place to
> go) for a csv/mysql export of MLBAM IDs mapped to retro/bbref IDs?
>
> I'll make a bookmark this time, I swear. Thanks.
>
> --
> Wells Oliver
> wells@...
>


---------------------------------------------
The Book--Playing The Percentages In Baseball
http://www.InsideTheBook.com

First email as part of the group.  Apologies if I am violating any protocols.

I've been trying to use the Baseball Databank (BDB) database to compute Win
Shares.  I've been following the book's step-by-step example with the 1998 St.
Louis Cardinals and I keep running into small discrepancies between the book and
the BDB.  Most of the time the book matches Baseball-Reference (B-R) and/or
Retrosheet.

Issues that I have encountered so far (all involved the 1998 NL):
1. Roberto Petagine (CIN) has 2 GIDP in BDB.  B-R has 1.  Retrosheet has him
hitting into two double plays, but one was a line drive type.
2. There are three pitcher HBP in the NL missing.  According to B-R John Thomson
(COL) has 2 and Gabe Gonzalez (FLO) has 1.  BDB has 0 for both.
3. The league fielding putouts (69710) to not match the total of the league
IPOuts (69719).
4. The league IPOuts (69719) do not match B-R (69720).  I haven't tracked down
where that stray out is yet.
5. For the entire NL, BDB has 837 WP, B-R has 835.  Again, I didn't track down
particular players yet.
6. The pitching runs allowed (not earned runs, just plain old runs) is off
significantly.  BDB has 11918, B-R 11943.  I did track down a couple of players
with Florida that have discrepancies.  Antonio Alfoseca has has 32 in BDB
(matching his ER) but 36 in B-R.  Felix Heredia has 25 in BDB (also matching his
ER) but 30 in B-R.  There are more discrepancies than those two though.

There are also issues with fielding assists, but that does not concern me as
much as the other stuff.

Is this the right place to report these issues?  I am willing to help track down
discrepancies, upon request.

Paul Golba

> Is this the right place to report these issues?  I am willing to help
> track down discrepancies, upon request.
>
> Paul Golba
>

This is absolutely the right place.  And while you didn't break any
protocols (if we even have any), from where I sit you can break whatever
protocols you need to, to report data issues like this.

Truth > whatever

Tom