TBDSG: Item Response Theory - Using data and models to recognize good questions

I would like to emphasize to everyone that the business address is not sufficient to find the door in the photo above. That's why I use the photo above.

I have entered the lat/lon and a gMaps link to the actual door pictured above in the "how to find us" section of this announcemnt. There is always plenty of free parking steps away from that door.

If you don't already know, behind that door you will find one of the Bay Area's most remarkable spaces. Join us for what is sure to be an intellectually stimulating discussion!

Agenda

We'll gather beginning at 6:30 in the Alessi Kitchen (La Cucina) for some food and non-alcoholic beverages. A little after 7 we'll convene in the dining room for the presentation followed by...?

The presentation will last for 45 minutes to an hour. After that we have several options for people that are able to stay:

1. Take a break and reconvene to continue the discussion as a group in the meeting room
2. Adjorn to the kitchen for small group discussions
3. Head to a bar :)

Daavid Stein's presentation:
Test Analysis with Item Response Theory Models: Learner Ability, Question Information & Test Reliability

Abstract

Item response theory is part of the field of psychometrics, concerned with the design and administration of questionnaires and tests. Using models and methods from IRT, we can learn which items or questions give us less information about the latent trait we want to measure (for instance, student subject knowledge). We will briefly go over the basics of a 2-parameter logistic models, generate some synthetic data, fit some models on that data and interpret the results. We’ll also take a look at why some standardized tests aren’t very good at differentiating students of different abilities and the impact this has on predicting student scores using supervised learning.

BIO

Daavid's LinkedIn Profile

Daavid Stein is a data scientist and machine learning engineer who enjoys working on problems for the betterment of humanity and the environment, such as understanding how to help students learn better or identifying invasive species from pictures taken by tree-planting drones. A one-time educator, Daavid is the former owner of MathBusters! Tutoring and taught a series of educational field trips at roller-skating rinks in Tampa. Daavid is available for new data science and machine learning projects at your business or organization.