Wed Apr 14

SIMON INITIATIVE RESEARCH PROJECT MSP Team: Jenny Mao, Jinghao Huang,
Rebecca Pei

sl 5 was merging needed/difficult?

sl 6 comparison of 2020 with 2019? [yup that's what they're doing]

overall: clustering within instructor or class? [MLM]

INSTR DATA

sl 9 differences in fall do not replicate to spring? [instr logins]

sl 10 what's going on at week 14? [COVID!]

sl 11 moving spikes in fall interesting

STUENT PERF DATA

sl 13 increase in high stakes performance in spr 20 interesting

  [instructors more rigorous in spr 20?]

sl 14 there seems to be a big overlap - means are different but lots
of distribution overlap

STUDENT LOGIN DATA

sl 16-17? seems to be a dropoff after COVID hits...

----------------

PPS -- Retention/Mobility MSP Team: Gloria Guo, Jenny Luo, Yuhang Ying

SLIDE NUMBERS!!!

(we don't remember what the 1st and 2nd research question is... give
us reminders or give them better names)

EDA:

initial EDA plot interesting; inequality in promise but not in use?

   vertical axis not counts, what are they?

   can we track non-PA enrollemnt?

next slide - same effects for gender, with male disadvantaged

(sl 11) nex slide: CTE effects but not free lunch effects

Methods:

sl 14 can you look at differnces in college type?

Results:

GPA and attendance are criteria for promise, right?  so ....

** keystone mean effect is very small even if significant

race is nearly significant; not gender

sl 17 big differences vetween 2018 and 2019 [retention, but consistent
across several measures]

sl 18 is there a smaple size problem for "yes" in 2019

RQ2 method 1 (regr disc) seems really interesting

1700 and 1300 seem like a lot,.. why are you worried about sample size




NBA Research Project (Basketball) MSP Team: Andrew Liu, Reed Peterson,
Willis Lu


sl 7: 

Have you considered incorporating other shift players' +/- in
adjusting this player's +/-?

sl 9:

ridge and lasso can be reinterpreted as bayes in which case posterior
std errors would be available

sl 10:

nested ridge regression is a lot like the multilevel models we looked
at last semester

sl 12 high means go with high se's

sl 14 regress on team rating?

  how is team incorporated in the rookie model?