2,655 reputation
620
bio website camdp.com
location Ottawa
age
visits member for 1 years, 10 months
seen 4 hours ago

ex-quant doing data @shopify. We're hiring data analysts, scientists and statisticians!


Apr
20
answered Inference with only left-censored data
Apr
20
comment How to deal with multicollinearity without dropping variables
Adding an L2 penalizer will help control high multilinearity without dropping any variables. That's my preferred technique.
Apr
19
comment Inference with only left-censored data
Hmm, I mentioned $T \in [0,5]$ because the subject was only susceptible after being born ($t=0$). Am I interpreting left-censorship wrong? Is there no natural lower-bound?
Apr
19
comment What kind of distribution is this via t-distribution
So perfect you should write it as an answer yourself!
Apr
19
comment What kind of distribution is this via t-distribution
Think about this, what values can $X$ take on? Can $X$ be negative? Can $X$ be a non-integer?
Apr
19
asked Inference with only left-censored data
Apr
16
comment Monte carlo Finance Project
Not the right place to ask this: a Master's student should be overflowing with ideas.
Apr
14
answered When to terminate the Bayesian A/B test?
Apr
8
comment Comparing Bernoulli means across subpopulations in which the number of observed successes may be zero
All my priors are Uniform(0,1) (i.e. Beta(1,1) ). The 10000 is the number of draws from the posterior I perform, so I can do monte carlo estimates of the $P(p>p_i)$
Apr
6
answered Comparing Bernoulli means across subpopulations in which the number of observed successes may be zero
Apr
5
comment Training, testing, validating in a survival analysis problem
The take-away I found from this, and why I originally was attracted to this paper, was how to deal with censorship in survival predictions, i.e. what loss function to use (though rereading your question, you may not have censorship).
Apr
5
answered Training, testing, validating in a survival analysis problem
Apr
1
answered Fitting logistic function with pymc
Apr
1
comment Switchpoint detection with probabilistic programming (pymc)
For example, fit the model: $\lambda_1 p + \lambda_2 (1-p)$, where $p = 1/(1 + exp(-\beta t) )$? That would work I believe, and would allow for smooth transitions. You are correct that inference on $\beta$'s slope could determine if a switch point exists. I really like this, you should explore it more.
Apr
1
revised Switchpoint detection with probabilistic programming (pymc)
deleted 15 characters in body
Apr
1
comment Converting a parametric survival model to a cash flow model. How do I account for aging in the population?
correct me if I am wrong: you population consists only of widgets that lasted for 1000+ days. Widgets that failed before then are not included, correct? If a widget has survived for 100 additional days, by this you mean the widget has survived for 1100 days, and you would like to construct a new survival curve for them. Or do you mean they enter the study at 1100 days?
Apr
1
answered Switchpoint detection with probabilistic programming (pymc)
Mar
29
comment What is the right attitude toward open source machine learning toolkits?
My view is that good OS libraries should have sane, override-able, defaults. Scikit-Learn is a good example of this (Weka is poorer example). That way the programs are explicit only when required. To program your own just to have more control can be overkill in 90% of cases -- better is to fork to OS library and added in the pieces you want.
Mar
29
comment Good resources on Aalen additive models (survival)
All of these are behind some kind of pay wall. Can you provide alternative links or resources?
Mar
29
awarded  Investor