Player Graph is Not Displayed for Some Players Due to Error on Ratings Endpoint

[vash3g]

Player has no graph so far, played in same number of tournaments as other club members.
No graph: http://agagd.usgo.org/player/17440/

Has graph, same time period, same number of games
http://agagd.usgo.org/player/22183/

[leeschumacher]

xhr request to http://agagd.usgo.org/ratings/17440/ returns:
[
{
"elab_date": null,
"sigma": 0.0,
"rating": 0.0
},
{
"elab_date": "2015-10-24",
"sigma": 1.0425,
"rating": -2.0423800000000001
},
{
"elab_date": "2016-01-09",
"sigma": 0.75205,
"rating": 2.9282699999999999
},
{
"elab_date": "2016-02-13",
"sigma": 0.56854000000000005,
"rating": 2.8661699999999999
}
]

that null date in the first entry causes the graph render to fail. The script could protect itself against that, but the data should also be cleaned up.

[JonathanBresler]

The database table, Ratings, contains ratings and sigmas generated by the
C++ program that we currently use as well as ratings and sigmas that were
generated by Paul Matthews program. For the previous progra, we do not
have dates or the tournament that was used to calculate the ratings and
sigmas. Indeed, during that period, ratings were run once a month or once
a quarter so that there was not a mapping between ratings and individual
tournaments. As a result, the Tournament_Code for those ratings and
sigmas. Consequently the dates are '0000-00-00', a date that does not
exist and the Tournament_Code is NULL. These are the equivalent of NaN in
numerical calculations.

If you know of a way that we can "clean up the data" I would be interested
in hearing it. Perhaps we could then do so.

Jonathan

mysql> select * from ratings where Pin_Player = 17440 order by Elab_Date ;
+------------+----------+---------+------------+-----------------+
| Pin_Player | Rating | Sigma | Elab_Date | Tournament_Code |
+------------+----------+---------+------------+-----------------+
| 17440 | 0.00000 | 0.00000 | 0000-00-00 | NULL |
| 17440 | -2.04238 | 1.04250 | 2015-10-24 | roch20151024 |
| 17440 | 2.92827 | 0.75205 | 2016-01-09 | ngc20160109 |
| 17440 | 2.86617 | 0.56854 | 2016-02-13 | roch20160213 |
+------------+----------+---------+------------+-----------------+
4 rows in set (0.00 sec)

mysql>

On Sat, 19 Mar 2016, Lee wrote:

xhr request to http://agagd.usgo.org/ratings/17440/ returns:
[
{
"elab_date": null,
"sigma": 0.0,
"rating": 0.0
},
{
"elab_date": "2015-10-24",
"sigma": 1.0425,
"rating": -2.0423800000000001
},
{
"elab_date": "2016-01-09",
"sigma": 0.75205,
"rating": 2.9282699999999999
},
{
"elab_date": "2016-02-13",
"sigma": 0.56854000000000005,
"rating": 2.8661699999999999
}
]

that null date in the first entry causes the graph render to fail. The script could protect itself against that, but the data should also be cleaned up.

---

You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
#42 (comment)

[leeschumacher]

So, i'm not familiar with the code that generates the ratings/sigma from the data, that record doesn't look like it has a meangingful rating or sigma value either so just deleting it would solve their problem. The other approaches would be to fix the code that generates the json to filter out those values (since without a meaningful date you couldn't graph it anyway), and/or fix the js to ignore those values.

[JonathanBresler]

Let's focus for a moment on a specific example.
Which player's information should we look at to see the issue.

Jonathan

On Sat, 26 Mar 2016, Lee wrote:

So, i'm not familiar with the code that generates the ratings/sigma from the data, that record doesn't look like it has a meangingful rating or sigma value either so just deleting it would solve their problem. The other approaches would be to fix the code that generates the json to filter out those values (since without a meaningful date you couldn't graph it anyway), and/or fix the js to ignore those values.

---

You are receiving this because you commented.
Reply to this email directly or view it on GitHub:
#42 (comment)

[leeschumacher]

How about http://agagd.usgo.org/ratings/17440/ ?

[michaelhiiva]

It seems the code responsible for this is not checking invalid dates i.e., null, or 0000-00-00

agagd/agagd/agagd_core/views/core.py

Lines 86 to 105 in 3276d25

	def member_ratings(request, member_id):
	#returns a members rating data as a json dict for graphing
	try:
	player = Member.objects.get(pk=member_id)
	ratings = player.ratings_set.all().order_by('elab_date')
	ratings_dict = [
	{'sigma': r.sigma,
	'elab_date': r.elab_date,
	'rating': r.rating} for r in ratings
	if r.elab_date != None]
	if len(ratings_dict) <= 1:
	logger.debug('Ratings error: only one rating')
	return JsonResponse({
	'result': 'error',
	'message': 'Only one rated game. Not enough data'
	})
	return JsonResponse(ratings_dict)
	except:
	logger.debug('Ratings error', exc_info=1)
	return JsonResponse({'result':'error'})

[michaelhiiva]

Both players now have graphs:

https://agagd.usgo.org/player/17440/

http://agagd.usgo.org/player/22183/

However, for older ratings we are just seeing a json error. ie.,

https://agagd.usgo.org/ratings/7/

{
  "result": "error"
}

https://agagd.usgo.org/player/7/

https://agagd.usgo.org/ratings/1/

{
  "result": "error"
}

https://agagd.usgo.org/player/1/

In these cases, I think we should have the graph indicate that there is some issue with creation or that there is not enough data to display a graph.