Collaborate with Local Schools

[nikdata]

I'm not sure how we would go about doing this, but maybe we could partner up with a local school and create some interesting insights/dashboards for teachers and/or students (thinking out loud here):

• identifying topics that need more attention either by course, grade, student, etc.

• identifying students who may be at risk of poor performance and giving an opportunity to teachers to step in and assist

• help administrators understand what topics or courses that students find most interesting or least interesting and why

[sctyner]

Nik that's a great idea. Do you have any connections with Des Moines Public Schools? (Friends or relatives or coworkers' spouses who are teachers?)

[nikdata]

@sctyner Unfortunately I do not. I moved to the Des Moines area fairly recently and haven't had an opportunity to connect with any of the school systems. I wonder if anyone attending the conference has some contacts? Do you know if there are schools in the Ames area that may benefit. Or should we limit to Des Moines area only?

[sctyner]

I can send an email out to participants. I've also reached out to a friend who works in the system to get her thoughts. In my limited understanding (I don't have kids), the Ames public schools are the best performing in the state, and DSM schools would probably benefit more from our help, but Ames would be easier for me to find connections as many of my colleagues have children in Ames schools.

[lebebr01]

I have at least one connection with an individual on the Assessment, Data, and Evaluation team and also have passively met their director. I can reach out to them, but I'd guess they would be hesitant providing data due to FERPA concerns. May be able to get aggregate school level data, however this may be less interesting due to being a single school district.

In a quick look at their website, it also looks like they use tableau for some visualizations: https://public.tableau.com/profile/des.moines.public.schools5887#!/

[mkosmicki]

DMPS and Johnston both use Tableau for visualization. The only reason I know is because they attend the Tableau Users Group for DSM. I moved here with no kids in tow, so I don't really have any connections either. Would be be able to get anonymized data from the state dept of ed?

[sctyner]

https://data.iowa.gov/browse?category=Education looks promising!

[DrFletch]

I work in the Division of Community Colleges and Workforce Preparation (includes Career and Technical Education that spans secondary & post-secondary) at the Iowa Department of Education. I feel there are numerous reporting and analysis opportunities that could be enhanced using RStudio. Possibly more promising opportunities can be found here: https://educateiowa.gov/adult-career-and-community-college/publications

[sctyner]

I have contacted some folks in the DMPS system, so we'll see where that goes!

[DrFletch]

Additional consumer reports down to the "school-level" about Iowa's K-12 systems can also be found via [(https://educateiowa.gov/data-reporting/district-aea-reports)] [(https://educateiowa.gov/education-statistics)]

I spoke to a couple of individuals and student-level data cannot be shared due to FERPA concerns...

[mharmston]

Should a group pursue a project associated with educational institutions, I would find such a direction particularly interesting. @DrFletch , does the FERPA concern generalize to cases where student-level data has been stripped of PII?

[linkalis]

I'd also be interested in working on a topic related to educational data. I imagine most school systems don't have data that would be deidentified enough that they could readily share, but I googled around a bit for "educational data mining" datsets and stumbled across these two that seem to be floating around:

• https://www.kaggle.com/aljarah/xAPI-Edu-Data/version/6

• https://archive.ics.uci.edu/ml/datasets/student+performance

I've recently heard about some research that takes methods designed for recommender systems (think: Netflix, Amazon recommendations) and applies them to predicting student grades/performance in courses. I don't have a huge math background, so I can't claim to know much about the specifics of these algorithmic approaches, but I know there are matrix factorization techniques to extract "factors" that describe students and courses, and uses those to predict student performance for new student/course combinations. I'd love to try to dig into recommender systems algorithms a bit more, try to implement something small-scale in R, and think how these approaches might be useful to education and beyond. (I know that recommender systems are already being used widely in marketing, so I'd love to think about how these algorithms could be purposed for education, human services, or other fields.)

Here's an article that's a nice example of some of the research I've heard of: https://epubs.siam.org/doi/abs/10.1137/1.9781611974973.62

And it looks like there are a few R libraries for recommender systems that might help us get started: https://gist.github.com/talegari/77c90db326b4848368287e53b1a18e8d