BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//cfp.jupytercon.com//EHPYEN
BEGIN:VTIMEZONE
TZID:Europe/Paris
BEGIN:STANDARD
DTSTART:20001029T030000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:CET
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20000326T020000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:CEST
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-2023-EHPYEN@cfp.jupytercon.com
DTSTART;TZID=Europe/Paris:20230512T140000
DTEND;TZID=Europe/Paris:20230512T160000
DESCRIPTION:This is part 2 of a two part tutorial. It is not recommended to
attend Part 2 without having attended Part 1.\n\nTutorial notebooks:\n\n-
https://github.com/soda-inria/survival-analysis-benchmark\n\n\nHere is th
e agenda of for the full session:\n\nAccording to Wikipedia:\n\nSurvival a
nalysis is a branch of statistics for analyzing the expected duration of t
ime until one event occurs\, such as deaths in biological organisms and fa
ilure in mechanical systems. [...]. Survival analysis attempts to answer c
ertain questions\, such as what is the proportion of a population which wi
ll survive past a certain time? Of those that survive\, at what rate will
they die or fail? Can multiple causes of death or failure be taken into ac
count? How do particular circumstances or characteristics increase or decr
ease the probability of survival?\n\nIn this two-part tutorial (morning an
d afternoon)\, we will deep dive into a practical case study of predictive
maintenance using tools from the scientific Python ecosystem. Here is a t
entative agenda:\n\nPart 1 (Morning)\n- What is time-censored data and why
it is a problem to train time-to-event regression models.\n- Single event
survival analysis with Kaplan-Meier using scikit-survival.\n- Competing r
isks modeling with Nelsonâ€“Aalen\, Aalen-Johansen using lifelines.\n- Eva
luation of the calibration of survival analysis estimators using the integ
rated brier score (IBS) metric.\n- Predictive survival analysis modeling w
ith Cox Proportional Hazards\, Survival Forests using scikit-survival\, Gr
adientBoostedIBS implemented from scratch with scikit-learn.\n- Estimation
of the cause-specific cumulative incidence function (CIF) using our Gradi
entBoostedIBS model.\n\nPart 2 (Afternoon)\n- How to use a trained Gradien
tBoostedIBS model to estimate the median survival time and the probability
of survival at a fixed time horizon.\n- Measuring the statistical associa
tion between input features and survival probabilities using partial depen
dence plot and permutation feature importance.\n- Presentation of the resu
lts of a benchmark of various survival analysis estimators on the KKBox da
taset.\n- Extracting implicit failure data from operation logs using sessi
onization with Ibis and DuckDB.\n- Hands-on wrap-up exercise.\n\nTarget au
dience: good familiarity with machine learning concepts\, with prior exper
ience using scikit-learn (you know what cross-validation means and how to
fit a Random Forest on a Pandas dataframe).
DTSTAMP:20231003T204124Z
LOCATION:Room 3 (Tutorial)
SUMMARY:Predictive survival analysis and competing risk modeling with sciki
t-learn\, scikit-survival\, lifelines\, Ibis\, and DuckDB (Part 2) - Guill
aume Lemaitre\, Vincent Maladiere\, Olivier Grisel
URL:https://cfp.jupytercon.com/2023/talk/EHPYEN/
END:VEVENT
END:VCALENDAR