Learning Analytics – Big Data

Day 1 – Introduction

  1. If you have data and a pattern you can predict the further data probabilistically
  2.  Examples
    1. Bordeaux
      1. Originally experts would taste the wine during the move from the barrel to the bottle
      2. Ashenfelter’s approach – He took price as a proxy for quality and historical data.
        1. Weather – poorly understood, but widely believed to affect wine quality
        2. Wine quality is a function of winter rain, growing temperature and harvest rain.
        3. What is the role of theory? what is the role of data?
          1. You need theory to help you find the data to look at. The data here might be proxy for something else.
          2. In the literature it seems like this is highly a-theoretical, but the data that show up to the part are a function of the data you invite to the part. The data you invite to the part are a function of how you see the world.
        4. The grape growers and the wine tasters reacted poorly.
          1. the mystique is gone
          2. the special knowledge about wine growing was taken away
          3. it takes the art out of wine making
      3. Could it be that the educational endeavor is like wine tasting? Do we want to focus on the art and keep the science out of it. – Definitely Maybe
        1. Admissions – who is likely to succeed
        2. Design – predict which approach will support engagement, satisfaction, learning.
          1. Andy says: Design is a process of decisions making. Someone who is designing will try to predict how well something is going to work.
  3. How would you study student engagement in videos?
    1. Define the construct (engaging), decide what constitutes an instructional video
    2. Collect and analyze data
  4. Edx findings
    1. engagement stops at 6 minutes
    2. informal are more engaging
    3. speaking fairly fast with high enthusiasm
    4. khan academy better than powerpoint
    5. interspersing talking head with tablet drawings is better than either alone.
  5. Summary
    1. Learning analytics it NOT a-theoretical
    2. There is an iterative loop between the theory and the data, they can both drive each other
    3. Predictive power is useful when ____
      1. something has to come along behind predictions

Day 2 – the difference between LA and educational research

  1. ancestry.com – LA has grown out of the .com movement
  2. focus on data exhaust – generated as a by-product of something else you’re doing
    1. Emission standards – what kinds of data do students emit
      1. metacognitive – check grades, schedule, advance, what time of day, what part of the semester
      2. discussions – reading posts, responding, length
      3. likes and dislikes
      4. IP address – location (on/off campus), type of network, device (mobile, computer, OS, browser)
      5. path – where you came from, where you go next
      6. assessment data
      7. how much of a video, where watched.
      8. Inside the institution – LMS, registrar, zip code
      9. Outside the institution – likes, shares
  3. Ethics – do no harm
    1. making unfounded assumptions
    2. must disclose and get permission to collect/use data
    3. using information that students don’t mean to share with you in the context of this class
  4. Who’s working here
    1. SOLAR – Society for LA Research – this one is less technical
    2. International Educational Data Mining
  5. What kind of work
    1. get generalizations from survey data that can’t be collected regularly
    2. find behaviors that predict success, red flags
    3. interacting with teacher, tutor, others
    4. How well are they using the LMS, or how easy is the LMS for them to use
    5. Where should you study? – connect behaviors to outcomes
    6. you have to look at the right data for the questions you are asking. You need to look at the assumptions that you are making as well.
  6. Examples
    1. Mcfayden & Dawson – Mining LMS data to develop early warning system for educators: a proof of concept
      1. SNA
      2. predictors of success
    2. Mendez, Ochoa, Chiluiza – techniques for data-drive curriculum analysis
      1. path through a degree program
      2. students dropping out before they take any CS classes.
Advertisements
This entry was posted in Speakers. Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

w

Connecting to %s