BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//wp-events-plugin.com//7.2.3.1//EN
TZID:America/New_York
X-WR-TIMEZONE:America/New_York
BEGIN:VEVENT
UID:907@fds.yale.edu
DTSTART;TZID=America/New_York:20260128T113000
DTEND;TZID=America/New_York:20260128T130000
DTSTAMP:20260115T153431Z
URL:https://fds.yale.edu/events/fds-colloquium-harsh-parikh/
SUMMARY:FDS Colloquium: Harsh Parikh (Yale)\, "Causal Inference Beyond Comm
 on Support"
DESCRIPTION:\nAbstract: Generalizing causal findings from randomized contro
 lled trials (RCTs) to target populations is a core challenge in modern sta
 tistics—one where violations of the positivity assumption force an uncom
 fortable choice between imprecise estimates and unverifiable parametric as
 sumptions. This talk presents two complementary methodological contributio
 ns that reframe how we diagnose and address these violations. First\, we i
 ntroduce the Rashomon Set of Optimal Trees (ROOT)\, a novel functional opt
 imization framework that leverages the multiplicity of near-optimal decisi
 on trees to interpretably characterize underrepresented subgroups. ROOT id
 entifies covariate regions contributing most to estimator variance\, enabl
 ing transparent communication of where reliable inference ends—transform
 ing a nuisance problem into actionable diagnostics. Second\, we develop a 
 unified regularization framework for extrapolation that replaces hard non-
 negativity constraints on estimator weights with a tunable soft penalty. T
 his formulation reveals a novel "bias-bias-variance" tradeoff—explicitly
  balancing covariate imbalance bias\, model misspecification bias\, and va
 riance—yielding a continuum of estimators with provable properties for n
 avigating extrapolation risk. We demonstrate both methods through a case s
 tudy generalizing the START trial (comparing opioid agonist therapies) to 
 a national treatment population. Together\, these tools offer statistician
 s a principled workflow: first characterize the boundaries of credible inf
 erence\, then carefully extend them through regularization when necessary.
 \n\n\n\nSpeaker Bio: My research focuses on developing (interpretable) cau
 sal inference approaches for aiding decisions in high-stakes complex scena
 rios. My collaborators and I have used my research to address challenges i
 n healthcare\, public health\, and social sciences. Decision-making in the
 se critical domains is fraught with difficulties stemming from\, but not l
 imited to\, the intricate interplay of factors\, including the heterogenei
 ty of causal effects across subpopulations\, the substantial costs associa
 ted with suboptimal decisions\, and the inherent complexities in the avail
 able data\, all of which complicate the assessment of risk-benefit trade-o
 ffs. In pursuit of more effective solutions\, my work is centered around t
 he development of causal inference methodologies that are:\n\n\n\nAccurate
 : to ensure accurate estimation of heterogeneous causal effects\, even in 
 scenarios with data limitations\, offering decision-makers a reliable foun
 dation upon which to base their choices.\n\n\n\nTrustworthy: to empower do
 main experts to comprehend the inner workings of the causal inference proc
 ess. This not only enables experts to validate the underlying assumptions 
 but also guarantees patients' safety.\n\n\n\nDomain-conscious: to bridge t
 he research-to-practice gap and yield solutions that are readily implement
 able. I leverage the context and domain knowledge to tailor solutions spec
 ific to a subject matter.\n
CATEGORIES:Fellows Events,FDS Events,Colloquium
LOCATION:Yale Institute for Foundations of Data Science & Webcast\, 219 Pro
 spect Street\, 13th Floor\, New Haven\, CT\, 06511\, United States
X-APPLE-STRUCTURED-LOCATION;VALUE=URI;X-ADDRESS=219 Prospect Street\, 13th 
 Floor\, New Haven\, CT\, 06511\, United States;X-APPLE-RADIUS=100;X-TITLE=
 Yale Institute for Foundations of Data Science & Webcast:geo:0,0
END:VEVENT
BEGIN:VTIMEZONE
TZID:America/New_York
X-LIC-LOCATION:America/New_York
BEGIN:STANDARD
DTSTART:20251102T010000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
END:STANDARD
END:VTIMEZONE
END:VCALENDAR