AssertionError: The key 'team' does not exist for entity 'Game'.

[roodhouse]

Hi, it's me again.

I wasn't scared off. I am trying to work my way through the wiki page.

Following the examples from "an introduction to the query interface" I type in this:

import nfldb
db = nfldb.connect()

q = nfldb.Query(db)
q.game(season_year=2013, season_type='Regular', team='NE', week=1)
for d in q.as_drives():
print d

And this error returns:

Traceback (most recent call last):
File "C:\Users\Rugh\Desktop\nfldb\test.py", line 5, in
q.game(season_year=2013, season_type='Regular', team='NE', week=1)
File "C:\Python27\lib\site-packages\nfldb\query.py", line 586, in game
_append_conds(self._default_cond, types.Game, kw)
File "C:\Python27\lib\site-packages\nfldb\query.py", line 202, in _append_conds
% (kbare, entity.name)
AssertionError: The key 'team' does not exist for entity 'Game'.

Each time I try to use 'team' in q.game I get this message.

What am I missing?

[ochawkeye]

I don't think you're missing anything. I noticed the same issue in the sample code provided.

Try:

q = nfldb.Query(db)
q.game(season_year=2013, season_type='Regular', week=1)
q.play_player(team='NE')
for d in q.as_drives():
    print d

I know there is probably a better way there than q.play_player, but it seems to work. @BurntSushi can probably explain

What am I missing?

Github markdown FYI: If you add

#```python

and

#```

around your code (minus the pound signs) that you post here, it will make it pretty.

[BurntSushi]

Dammit. This is a genuine bug. I got overzealous with error checking in the last release. Nice find!

[ochawkeye]

I'm going to change my answer here - something appears to be wrong. Almost every example gives a q.game(team='NE') methodology...

[BurntSushi]

P.S. @ochawkeye Have you see the write up for the nfldb 0.2.0 release? https://github.com/BurntSushi/nfldb/releases/tag/0.2.0 (TL;DR I completely re-wrote nfldb.query.)

[BurntSushi]

OK, this is fixed.

@roodhouse You will need to run pip install --upgrade nfldb and then the examples should work. Sorry about that!

@ochawkeye P.S. Check out the fix. We have regression tests now. :-)

[roodhouse]

Thanks guys. Time to test my newly learned skill. If I come back here
then you know what happened..

On Wed, Jul 30, 2014 at 5:46 PM, Andrew Gallant [email protected]
wrote:

OK, this is fixed.

@roodhouse https://github.com/roodhouse You will need to run pip
install --upgrade nfldb and then the examples should work. Sorry about
that!

@ochawkeye https://github.com/ochawkeye P.S. Check out the fix. We have
regression tests now. :-)

—
Reply to this email directly or view it on GitHub
#34 (comment).

[ochawkeye]

Out of curiosity, how expensive is a database query as opposed to iterating over the data redundantly?

q = nfldb.Query(db)
q.game(season_year=2013, season_type='Regular', week=1, team='NE')
q.drive(pos_team='NE')
for d in q.as_drives():
    print d

q = nfldb.Query(db)
q.game(season_year=2013, season_type='Regular', week=1, team='NE')
q.drive(pos_team='BUF')
for d in q.as_drives():
    print d

vs.

q = nfldb.Query(db)
q.game(season_year=2013, season_type='Regular', week=1, team='NE')
for d in q.as_drives():
    if d.pos_team == 'NE':
        print d
for d in q.as_drives():
    if d.pos_team == 'BUF':
        print d

Running first sample 100 times:

Execution time for 100 cycles = 2.66303673806
Average per cycle = 0.0266303673806

Running second sample 100 times:

Execution time for 100 cycles = 4.58160582431
Average per cycle = 0.0458160582431

But repeating tests gives drastically varying times. Any thoughts on this?

[BurntSushi]

@ochawkeye As a general rule of thumb, the more you put into a database query and the less work you do in Python, the faster it will be. I think that is consistent with your tests because you're doing more work in Python in the second example.

I would suggest removing week=1 and doing the test over an entire season to get stabler numbers.

Also, your tests should not have print in them. That could also be a source of instability between test runs. Maybe do a sum of something instead.

(If you wait, I'll do it when I get home later today.)