Mon, 11/05/2012 - 7:04pm
Question from President Fuhrman: Human subjects, IRB issues, informed consent...
Dr. Baker responds: "It's impractictal... depends on risk and what the data is used for... there are no standards around this yet... in the corporate world, consent is very limited, and it is being used for things not in users' interests"
"Datashop data is anonymous; hard copies if people really want to connect the numbers"
Sharon Hsiao Says:Mon, 11/05/2012 - 7:01pm
EDM hasn't been looked to help learning disability students. Which should!!!
Demetri Lales Says:Mon, 11/05/2012 - 7:08pm
This is extremely surprising seeing as how there are so much educational apps that help students with special needs such as autism, dyslexia, ADHD, etc.
Janice Joo Says:Mon, 11/05/2012 - 7:04pm
That is a fantastic idea! If there is such an example, we need to write about it in NLT.
Janice Joo Says:Mon, 11/05/2012 - 6:59pm
Question from audience about how to make the data useful for the student, and not just the teacher: Dr. Baker's response is "open learner modeling", which was Sharon's thesis!
Sharon Hsiao Says:Mon, 11/05/2012 - 7:03pm
Yes!!
I have a long list of literature to share about open student modeling. Contact me you are interested in it.
Mon, 11/05/2012 - 6:57pm
Regarding effectiveness of EDM interventions
16% - is about the kid
55% - is about the curriculum/ game design
Did I hear this correctly?
If so, this means that the diagnosis of the kids is the lesser issue on outcomes
Sharon Hsiao Says:Mon, 11/05/2012 - 6:53pm
Pslc datashop has the standardized parameters for using the EDM methods. But a lot softwares out there still have to collect their own logs and data. Ryan gave the example, it's the process, like , autocorrection technique, originally handwriting recognition tools have to detect all those, but eventually, this field will emerge and find a more universal accessible model/method.
Janice Joo Says:Mon, 11/05/2012 - 6:52pm
Good examples of how to present data:
- Neil and Christina Heffernan
- "Produce Signals" Project
Curious to explore these examples
Janice Joo Says:Mon, 11/05/2012 - 6:48pm
Question from audience about Early Childhood Education applications; Dr. Baker responds briefly that iPads have enabled the collection of data from young users...
Demetri Lales Says:Mon, 11/05/2012 - 6:48pm
Part his work includes working with preschoolers and iPad apps for them.
Janice Joo Says:Mon, 11/05/2012 - 6:45pm
After school science programs: Dr. Bakers says there is an ongoing NSF project to try and find out which students COULD find success in STEM careers, but don't believe they can do so; I'm curious to hear which study that is!
Demetri Lales Says:Mon, 11/05/2012 - 6:42pm
Scooter the tutor a dog character that helps students and intervene when necessary. Helps lessen students who game the system.
Sharon Hsiao Says:Mon, 11/05/2012 - 6:40pm
Predict the future, is it enough?
Micro level: to do intervention in students learning. (To change he future)
Demetri Lales Says:Mon, 11/05/2012 - 6:39pm
students who do not self explain do not learn the material.
Sharon Hsiao Says:Mon, 11/05/2012 - 6:41pm
Stay tuned on NLT research digest, there will be a piece about the effects of self-explanation in learning study:)
Demetri Lales Says:Mon, 11/05/2012 - 6:38pm
EDM predicts which students are prepared for future learning. Students who are seek help when they need it. Also predicts which student are most likely to fail, and drop out of college. Is it enough to predict the future?
Janice Joo Says:Mon, 11/05/2012 - 6:38pm
"reason to predict the future is to change the future"
Sharon Hsiao Says:Mon, 11/05/2012 - 6:37pm
His edm approach, Collect field observation data combined log data, use data mining methods.
Not just predict the success for the problem, but to answer the higher question, success for next course? New skills? Career? Motivational factors?
Janice Joo Says:Mon, 11/05/2012 - 6:40pm
awesome! way to cover all the bases! we should consider the possibility of liveblogging vs. live-tweeting at EdLab.
Nara Kasbergen Says:Mon, 11/05/2012 - 6:45pm
Probably depends what you're after.
Liveblogging is a better way to record the experience for yourself and reference later.
Live-tweeting is a better way to connect with others who are having the same experience on the spot.
For something this small, live-tweeting is kind of pointless but for larger events it is definitely worthwhile. Pretty much all the contacts I made in my two years attending Games For Change, I made through live-tweeting and the conversations it spawned.
However, I do find live-tweeting anything to be good practice. It takes some practice to get good at distilling the salient points down to 140 characters (usually less with the hashtags) so I jump on any opportunity to try and improve!
Demetri Lales Says:Mon, 11/05/2012 - 6:35pm
Uses include showing teachers which students are engaged, frustrated, bored, etc
Janice Joo Says:Mon, 11/05/2012 - 6:35pm
Prediction and collection of student emotion data is improving quickly; currently it is only 25% better than chance
*student emotions such as frustration, engagement, positive/negative experience
Demetri Lales Says:Mon, 11/05/2012 - 6:33pm
Can detect students who game the system. Predicts new student behavior accurately.
Sharon Hsiao Says:Mon, 11/05/2012 - 6:28pm
Ryan gives some common EDM methods, knowledge assessment, structured discovery, mining relations. etc.
The some flavors of the benefits: detect off task behavior, wtf: without teaching facilitation behavior, careless errors, gaming etc.
Janice Joo Says:Mon, 11/05/2012 - 6:28pm
Various disengagement behaviors... among which Dr. Baker presents the "WTF Behavior"... lol
Mon, 11/05/2012 - 6:25pm
Synonyms: "Educational Data Mining" (EDM), Learning Analytics
Used for basic research and improving educational practice
i.e. Assisting the student WHILE assessing them. Kids do the homework as usual (just online and not on paper), while they receive help AND teachers get formative assessment data
Sharon Hsiao Says:Mon, 11/05/2012 - 6:21pm
Predicting future from past and present.
The power of EDM is to use automatic ways to try A lot of different models with a lot of different methods, to search the relationships among the data.
Janice Joo Says:Mon, 11/05/2012 - 6:17pm
"Data doesn't have to be collected; just connected"
Demetri Lales Says:Mon, 11/05/2012 - 6:17pm
Increasing amounts of data from educational software, and attitudes and career services are becoming available for analysis.
Janice Joo Says:Mon, 11/05/2012 - 6:16pm
Because of educational software, we are able to collect data about students' learning behavior: rich source of big data in education about learning and engagement
Most of that data used to be locked up, but now becoming more available, i.e. PSLC Data Shop
Sharon Hsiao Says:Mon, 11/05/2012 - 6:15pm
Educational softwares allow to collect more and more data other than just scores!
Demetri Lales Says:Mon, 11/05/2012 - 6:13pm
Data in class used to be on paper, valuable info for classrooms not recorded at all.
Sharon Hsiao Says:Mon, 11/05/2012 - 6:12pm
Today's talk is about Educational Datamining - predict future.
Big data -> data mining.
He gave himself as the example of using several products, browsing behavior, and those data has been used to sell more products to him.
Mon, 11/05/2012 - 6:33pm
Introductions by Provost Dr Thomas James. President Susan Fuhrman in front row
Source: http://edlab.tc.columbia.edu/index.php?q=node/8474
c. difficile carmelo anthony nurse jackie nurse jackie peeps nhl playoffs masters
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.