Machine Learning Meetup Notes: 2010-05-23 - Revision history

SpammerHellDontDelete at 22:36, 28 May 2010

2010-05-28T22:36:13Z

← Older revision		Revision as of 15:36, 28 May 2010
Line 25:		Line 25:
	* + step success chance (% of successes total for this unique stepname)		* + step success chance (% of successes total for this unique stepname)
	* + student "IQ" (% successful answers by this student)		* + student "IQ" (% successful answers by this student)
	* + complexity (number of skills required for this ~~step~~)		* + complexity (number of skills required for this unique stepname)

SpammerHellDontDelete at 04:39, 27 May 2010

2010-05-27T04:39:51Z

← Older revision		Revision as of 21:39, 26 May 2010
Line 26:		Line 26:
	* + student "IQ" (% successful answers by this student)		* + student "IQ" (% successful answers by this student)
	* + complexity (number of skills required for this step)		* + complexity (number of skills required for this step)
	* + frequency of skills (e.g. discretize into low/medium/high frequency -- reasoning: infrequently tested skills may be harder/easier)

ThomasLotze at 03:52, 27 May 2010

2010-05-27T03:52:46Z

← Older revision		Revision as of 20:52, 26 May 2010
Line 2:		Line 2:

	We decided to remove (-) and add (+) the following features:		We decided to remove (-) and add (+) the following features:
	Row, (only used for submission, not for ML algorithms)		* Row, (only used for submission, not for ML algorithms)
	Anon Student Id,		* Anon Student Id,
	Problem Hierarchy,		* Problem Hierarchy,
	Problem Name,		* Problem Name,
	Problem View,		* Problem View,
	- Step Name,		* - Step Name,
	+ unique step name (step name+problem name)		* + unique step name (step name+problem name)
	- Step Start Time,		* - Step Start Time,
	- First Transaction Time,		* - First Transaction Time,
	- Correct Transaction Time,		* - Correct Transaction Time,
	- Step End Time,		* - Step End Time,
	- Step Duration (sec),		* - Step Duration (sec),
	- Correct Step Duration (sec),		* - Correct Step Duration (sec),
	- Error Step Duration (sec),		* - Error Step Duration (sec),
	Correct First Attempt,		* Correct First Attempt,
	- Incorrects,		* - Incorrects,
	- Hints,		* - Hints,
	- Corrects,		* - Corrects,
	- KC(...),		* - KC(...),
	- Opportunity(...)		* - Opportunity(...)
	+ set of superskills (either boolean or opportunity value) (superskills = clustered skills)		* + set of superskills (either boolean or opportunity value) (superskills = clustered skills)
	+ step success chance (% of successes total for this unique stepname)		* + step success chance (% of successes total for this unique stepname)
	+ student "IQ" (% successful answers by this student)		* + student "IQ" (% successful answers by this student)
	+ complexity (number of skills required for this step)		* + complexity (number of skills required for this step)
	+ frequency of skills (e.g. discretize into low/medium/high frequency -- reasoning: infrequently tested skills may be harder/easier)		* + frequency of skills (e.g. discretize into low/medium/high frequency -- reasoning: infrequently tested skills may be harder/easier)

SpammerHellDontDelete at 02:47, 27 May 2010

2010-05-27T02:47:30Z

← Older revision		Revision as of 19:47, 26 May 2010
Line 2:		Line 2:

	We decided to remove (-) and add (+) the following features:		We decided to remove (-) and add (+) the following features:
	- Row,		Row, (only used for submission, not for ML algorithms)
	Anon Student Id,		Anon Student Id,
	Problem Hierarchy,		Problem Hierarchy,
Line 23:		Line 23:
	- Opportunity(...)		- Opportunity(...)
	+ set of superskills (either boolean or opportunity value) (superskills = clustered skills)		+ set of superskills (either boolean or opportunity value) (superskills = clustered skills)
	+ step success chance (% of successes total for this ~~step~~)		+ step success chance (% of successes total for this unique stepname)
	+ student IQ (% successful answers)		+ student "IQ" (% successful answers by this student)
	+ complexity (number of skills required)		+ complexity (number of skills required for this step)
	+ frequency of skills (e.g. discretize into low/medium/high frequency)		+ frequency of skills (e.g. discretize into low/medium/high frequency -- reasoning: infrequently tested skills may be harder/easier)

SpammerHellDontDelete at 02:24, 27 May 2010

2010-05-27T02:24:24Z

← Older revision		Revision as of 19:24, 26 May 2010
Line 4:		Line 4:
	- Row,		- Row,
	Anon Student Id,		Anon Student Id,
	Problem Hierarchy,		Problem Hierarchy,
	Problem Name,		Problem Name,
	Problem View,		Problem View,
	- Step Name,		- Step Name,
	+ unique step name (step name+problem name)		+ unique step name (step name+problem name)
Line 16:		Line 16:
	- Correct Step Duration (sec),		- Correct Step Duration (sec),
	- Error Step Duration (sec),		- Error Step Duration (sec),
	Correct First Attempt,		Correct First Attempt,
	- Incorrects,		- Incorrects,
	- Hints,		- Hints,

SpammerHellDontDelete at 01:17, 25 May 2010

2010-05-25T01:17:40Z

← Older revision		Revision as of 18:17, 24 May 2010
Line 41:		Line 41:

	* We agreed that the accepted features above generally make sense to us; to define the set of features for the individual multi-algo problems is yet TODO.		* We agreed that the accepted features above generally make sense to us; to define the set of features for the individual multi-algo problems is yet TODO.



			Andy

SpammerHellDontDelete at 00:54, 25 May 2010

2010-05-25T00:54:29Z

← Older revision		Revision as of 17:54, 24 May 2010
Line 1:		Line 1:
	Erin, Theo and Andy came together to define KDD machine learning problem definition.		Erin, Theo and Andy came together to define KDD machine learning problem definition.

	We decided to remove (-~~), perhaps-remove (?~~) and add (+) the following features:		We decided to remove (-) and add (+) the following features:
	- Row,		- Row,
	Anon Student Id,		Anon Student Id,
	? Problem Hierarchy,		Problem Hierarchy,
	? Problem Name,		Problem Name,
	? Problem View,		Problem View,
	- Step Name,		- Step Name,
	+ unique step name (step name+problem name)		+ unique step name (step name+problem name)

SpammerHellDontDelete at 16:34, 24 May 2010

2010-05-24T16:34:15Z

← Older revision		Revision as of 09:34, 24 May 2010
Line 5:		Line 5:
	Anon Student Id,		Anon Student Id,
	? Problem Hierarchy,		? Problem Hierarchy,
	? Problem Name,		? Problem Name,
	? Problem View,		? Problem View,
	- Step Name,		- Step Name,
	+ unique step name (step name+problem name)		+ unique step name (step name+problem name)
	- Step Start Time,		- Step Start Time,
	- First Transaction Time,		- First Transaction Time,
	- Correct Transaction Time,		- Correct Transaction Time,
	- Step End Time,		- Step End Time,
	- Step Duration (sec),		- Step Duration (sec),
	- Correct Step Duration (sec),		- Correct Step Duration (sec),
	- Error Step Duration (sec),		- Error Step Duration (sec),
	Correct First Attempt,		Correct First Attempt,
	- Incorrects,		- Incorrects,
	- Hints,		- Hints,
	- Corrects,		- Corrects,
	- KC(...),		- KC(...),
	- Opportunity(...)		- Opportunity(...)
	+ set of superskills (either boolean or opportunity value) (superskills = clustered skills)		+ set of superskills (either boolean or opportunity value) (superskills = clustered skills)
	+ step success chance (% of successes total for this step)		+ step success chance (% of successes total for this step)
	+ student IQ (% successful answers)		+ student IQ (% successful answers)
	+ complexity (number of skills required)		+ complexity (number of skills required)
	+ frequency of skills (e.g. discretize into low/medium/high frequency)		+ frequency of skills (e.g. discretize into low/medium/high frequency)

SpammerHellDontDelete at 16:33, 24 May 2010

2010-05-24T16:33:59Z

← Older revision		Revision as of 09:33, 24 May 2010
Line 2:		Line 2:

	We decided to remove (-), perhaps-remove (?) and add (+) the following features:		We decided to remove (-), perhaps-remove (?) and add (+) the following features:
	- Row,		- Row,
	Anon Student Id,		Anon Student Id,
	? Problem Hierarchy,		? Problem Hierarchy,
	? Problem Name,		? Problem Name,
	? Problem View,		? Problem View,

SpammerHellDontDelete: Created page with 'Erin, Theo and Andy came together to define KDD machine learning problem definition. We decided to remove (-), perhaps-remove (?) and add (+) the following features: - Row, Anon…'

2010-05-24T16:33:48Z

Created page with 'Erin, Theo and Andy came together to define KDD machine learning problem definition. We decided to remove (-), perhaps-remove (?) and add (+) the following features: - Row, Anon…'

New page

Erin, Theo and Andy came together to define KDD machine learning problem definition.

We decided to remove (-), perhaps-remove (?) and add (+) the following features:
- Row,
Anon Student Id,
? Problem Hierarchy,
? Problem Name,
? Problem View,
- Step Name,
+ unique step name (step name+problem name)
- Step Start Time,
- First Transaction Time,
- Correct Transaction Time,
- Step End Time,
- Step Duration (sec),
- Correct Step Duration (sec),
- Error Step Duration (sec),
Correct First Attempt,
- Incorrects,
- Hints,
- Corrects,
- KC(...),
- Opportunity(...)
+ set of superskills (either boolean or opportunity value) (superskills = clustered skills)
+ step success chance (% of successes total for this step)
+ student IQ (% successful answers)
+ complexity (number of skills required)
+ frequency of skills (e.g. discretize into low/medium/high frequency)

== Submission datasets naming convention: ==
* "bridge"
* "algebra" -> also has KC rules model (need to orthoganlize them as well TODO erin)

== Multi-algorithm idea: ==
* After we discussed the features above, we had the idea to use multiple algorithms for predicting out output variable (probability of success for first try for this student-step): one algorithm predicting student success, one for step difficulty, perhaps additional ones... then have an aggregate function learn the overall success probability. Since we have about 3000 steps per student, we should have enough data to train a model for each student.

* We agreed that the accepted features above generally make sense to us; to define the set of features for the individual multi-algo problems is yet TODO.