andrew_gelman_stats andrew_gelman_stats-2013 andrew_gelman_stats-2013-2089 knowledge-graph by maker-knowledge-mining
Source: html
Introduction: The Stan meeting today reminded me of Joel Spolsky’s recasting of the Yiddish joke about Shlemiel the Painter. Joel retold it on his blog, Joel on Software , in the post Back to Basics : Shlemiel gets a job as a street painter, painting the dotted lines down the middle of the road. On the first day he takes a can of paint out to the road and finishes 300 yards of the road. “That’s pretty good!” says his boss, “you’re a fast worker!” and pays him a kopeck. The next day Shlemiel only gets 150 yards done. “Well, that’s not nearly as good as yesterday, but you’re still a fast worker. 150 yards is respectable,” and pays him a kopeck. The next day Shlemiel paints 30 yards of the road. “Only 30!” shouts his boss. “That’s unacceptable! On the first day you did ten times that much work! What’s going on?” “I can’t help it,” says Shlemiel. “Every day I get farther and farther away from the paint can!” Joel used it as an example of the kind of string processing naive programmers ar
sentIndex sentText sentNum sentScore
1 On the first day he takes a can of paint out to the road and finishes 300 yards of the road. [sent-3, score-0.668]
2 The next day Shlemiel only gets 150 yards done. [sent-7, score-0.416]
3 150 yards is respectable,” and pays him a kopeck. [sent-9, score-0.329]
4 The next day Shlemiel paints 30 yards of the road. [sent-10, score-0.339]
5 “Every day I get farther and farther away from the paint can! [sent-17, score-0.428]
6 The reason I bring it up is that software development almost inevitably employs the Shlemiel the Painter algorithm . [sent-19, score-0.354]
7 Here’s the problem in a nutshell: The more moving pieces your software has, the longer it takes to add a new feature or change an existing feature. [sent-20, score-0.684]
8 For example, when we have N special functions defined, if we want to change the way error handling works in all of them, it takes N units of work. [sent-21, score-0.496]
9 So the first feature takes one unit of time, the second two units, and so on. [sent-23, score-0.324]
10 And we all know where this goes, though if you’re like me rather than like Gauss, you didn’t derive the result in your head in primary school : The upshot is that to add features takes time proportional to . [sent-24, score-0.432]
11 In an ideal world, you look into the future when designing the algorithm the first time and imagine all the ways it might change and design something simple with that in mind. [sent-29, score-0.526]
12 At least that’s how software design works in theory. [sent-30, score-0.297]
13 It’s the unknown unknowns that get you every time. [sent-32, score-0.465]
14 But there are also unknown unknowns – there are things we do not know we don’t know. [sent-35, score-0.457]
15 It’s the unknown unknowns that get you every time. [sent-36, score-0.465]
16 One issue is that those without a lot of experience in software development can see refactoring or trying to design modularly in the first place as akin to rearranging the deck chairs on the Titanic . [sent-40, score-0.468]
17 But what it’s really about is trying to wrestle software into a manageable state. [sent-41, score-0.318]
18 Just an example related to Stan — we’re about to go through and rewrite everyone one of our distribution functions yet again so that we can take higher-order derivatives (we need this for some optimization, for Laplace approximations, and for RHMC). [sent-42, score-0.272]
19 We had basic tests, then we needed tests for vectorization, then tests for all the varying ways the functions could be called with data and parameters. [sent-50, score-0.404]
20 And at one point, we actually simplified all the distributions (again requiring N units of work) to get rid of the traits-based error-handling configuration that we anticipated needed but never needed. [sent-56, score-0.311]
wordName wordTfidf (topN-words)
[('shlemiel', 0.362), ('unknowns', 0.249), ('yards', 0.23), ('joel', 0.23), ('software', 0.19), ('rewrite', 0.163), ('painter', 0.145), ('takes', 0.141), ('unknown', 0.134), ('units', 0.131), ('distributions', 0.117), ('tests', 0.116), ('change', 0.115), ('paint', 0.111), ('vectorization', 0.111), ('functions', 0.109), ('day', 0.109), ('design', 0.107), ('feature', 0.106), ('farther', 0.104), ('pays', 0.099), ('development', 0.094), ('added', 0.091), ('stan', 0.086), ('proportional', 0.086), ('simple', 0.082), ('every', 0.082), ('work', 0.079), ('gets', 0.077), ('first', 0.077), ('time', 0.075), ('things', 0.074), ('fast', 0.071), ('re', 0.071), ('algorithm', 0.07), ('adding', 0.068), ('features', 0.068), ('existing', 0.066), ('dotted', 0.066), ('cholesky', 0.066), ('manageable', 0.066), ('modularity', 0.066), ('retold', 0.066), ('rhmc', 0.066), ('moving', 0.066), ('needed', 0.063), ('defined', 0.062), ('upshot', 0.062), ('modular', 0.062), ('wrestle', 0.062)]
simIndex simValue blogId blogTitle
same-blog 1 0.99999988 2089 andrew gelman stats-2013-11-04-Shlemiel the Software Developer and Unknown Unknowns
Introduction: The Stan meeting today reminded me of Joel Spolsky’s recasting of the Yiddish joke about Shlemiel the Painter. Joel retold it on his blog, Joel on Software , in the post Back to Basics : Shlemiel gets a job as a street painter, painting the dotted lines down the middle of the road. On the first day he takes a can of paint out to the road and finishes 300 yards of the road. “That’s pretty good!” says his boss, “you’re a fast worker!” and pays him a kopeck. The next day Shlemiel only gets 150 yards done. “Well, that’s not nearly as good as yesterday, but you’re still a fast worker. 150 yards is respectable,” and pays him a kopeck. The next day Shlemiel paints 30 yards of the road. “Only 30!” shouts his boss. “That’s unacceptable! On the first day you did ten times that much work! What’s going on?” “I can’t help it,” says Shlemiel. “Every day I get farther and farther away from the paint can!” Joel used it as an example of the kind of string processing naive programmers ar
2 0.13678305 1799 andrew gelman stats-2013-04-12-Stan 1.3.0 and RStan 1.3.0 Ready for Action
Introduction: The Stan Development Team is happy to announce that Stan 1.3.0 and RStan 1.3.0 are available for download. Follow the links on: Stan home page: http://mc-stan.org/ Please let us know if you have problems updating. Here’s the full set of release notes. v1.3.0 (12 April 2013) ====================================================================== Enhancements ---------------------------------- Modeling Language * forward sampling (random draws from distributions) in generated quantities * better error messages in parser * new distributions: + exp_mod_normal + gumbel + skew_normal * new special functions: + owenst * new broadcast (repetition) functions for vectors, arrays, matrices + rep_arrray + rep_matrix + rep_row_vector + rep_vector Command-Line * added option to display autocorrelations in the command-line program to print output * changed default point estimation routine from the command line to
3 0.13581926 1753 andrew gelman stats-2013-03-06-Stan 1.2.0 and RStan 1.2.0
Introduction: Stan 1.2.0 and RStan 1.2.0 are now available for download. See: http://mc-stan.org/ Here are the highlights. Full Mass Matrix Estimation during Warmup Yuanjun Gao, a first-year grad student here at Columbia (!), built a regularized mass-matrix estimator. This helps for posteriors with high correlation among parameters and varying scales. We’re still testing this ourselves, so the estimation procedure may change in the future (don’t worry — it satisfies detailed balance as is, but we might be able to make it more computationally efficient in terms of time per effective sample). It’s not the default option. The major reason is the matrix operations required are expensive, raising the algorithm cost to , where is the average number of leapfrog steps, is the number of iterations, and is the number of parameters. Yuanjun did a great job with the Cholesky factorizations and implemented this about as efficiently as is possible. (His homework for Andrew’s class w
4 0.1051648 2291 andrew gelman stats-2014-04-14-Transitioning to Stan
Introduction: Kevin Cartier writes: I’ve been happily using R for a number of years now and recently came across Stan. Looks big and powerful, so I’d like to pick an appropriate project and try it out. I wondered if you could point me to a link or document that goes into the motivation for this tool (aside from the Stan user doc)? What I’d like to understand is, at what point might you look at an emergent R project and advise, “You know, that thing you’re trying to do would be a whole lot easier/simpler/more straightforward to implement with Stan.” (or words to that effect). My reply: For my collaborators in political science, Stan has been most useful for models where the data set is not huge (e.g., we might have 10,000 data points or 50,000 data points but not 10 million) but where the model is somewhat complex (for example, a model with latent time series structure). The point is that the model has enough parameters and uncertainty that you’ll want to do full Bayes (rather than some sort
5 0.10123212 1627 andrew gelman stats-2012-12-17-Stan and RStan 1.1.0
Introduction: We’re happy to announce the availability of Stan and RStan versions 1.1.0, which are general tools for performing model-based Bayesian inference using the no-U-turn sampler, an adaptive form of Hamiltonian Monte Carlo. Information on downloading and installing and using them is available as always from Stan Home Page: http://mc-stan.org/ Let us know if you have any problems on the mailing lists or at the e-mails linked on the home page (please don’t use this web page). The full release notes follow. (R)Stan Version 1.1.0 Release Notes =================================== -- Backward Compatibility Issue * Categorical distribution recoded to match documentation; it now has support {1,...,K} rather than {0,...,K-1}. * (RStan) change default value of permuted flag from FALSE to TRUE for Stan fit S4 extract() method -- New Features * Conditional (if-then-else) statements * While statements -- New Functions * generalized multiply_lower_tri
6 0.090948328 2161 andrew gelman stats-2014-01-07-My recent debugging experience
7 0.09085907 1605 andrew gelman stats-2012-12-04-Write This Book
8 0.088048905 231 andrew gelman stats-2010-08-24-Yet another Bayesian job opportunity
9 0.087784022 1807 andrew gelman stats-2013-04-17-Data problems, coding errors…what can be done?
10 0.087650374 1580 andrew gelman stats-2012-11-16-Stantastic!
11 0.087605715 1710 andrew gelman stats-2013-02-06-The new Stan 1.1.1, featuring Gaussian processes!
12 0.085839324 515 andrew gelman stats-2011-01-13-The Road to a B
13 0.083355904 1514 andrew gelman stats-2012-09-28-AdviseStat 47% Campaign Ad
14 0.082574263 1948 andrew gelman stats-2013-07-21-Bayes related
15 0.081133574 2244 andrew gelman stats-2014-03-11-What if I were to stop publishing in journals?
16 0.080809541 2232 andrew gelman stats-2014-03-03-What is the appropriate time scale for blogging—the day or the week?
17 0.080641806 1520 andrew gelman stats-2012-10-03-Advice that’s so eminently sensible but so difficult to follow
18 0.079446815 1009 andrew gelman stats-2011-11-14-Wickham R short course
19 0.077767625 2150 andrew gelman stats-2013-12-27-(R-Py-Cmd)Stan 2.1.0
20 0.076113477 1596 andrew gelman stats-2012-11-29-More consulting experiences, this time in computational linguistics
topicId topicWeight
[(0, 0.183), (1, -0.001), (2, -0.027), (3, 0.048), (4, 0.079), (5, 0.024), (6, 0.045), (7, -0.077), (8, -0.023), (9, -0.08), (10, -0.079), (11, -0.032), (12, -0.012), (13, -0.056), (14, -0.004), (15, -0.01), (16, 0.009), (17, -0.004), (18, -0.007), (19, 0.014), (20, 0.017), (21, -0.0), (22, -0.028), (23, 0.02), (24, 0.016), (25, 0.003), (26, 0.001), (27, 0.019), (28, 0.031), (29, 0.005), (30, 0.023), (31, -0.033), (32, 0.042), (33, -0.004), (34, 0.029), (35, -0.055), (36, -0.039), (37, 0.065), (38, 0.004), (39, 0.016), (40, 0.001), (41, 0.009), (42, -0.054), (43, -0.002), (44, -0.013), (45, 0.071), (46, -0.044), (47, 0.012), (48, 0.057), (49, 0.014)]
simIndex simValue blogId blogTitle
same-blog 1 0.95871085 2089 andrew gelman stats-2013-11-04-Shlemiel the Software Developer and Unknown Unknowns
Introduction: The Stan meeting today reminded me of Joel Spolsky’s recasting of the Yiddish joke about Shlemiel the Painter. Joel retold it on his blog, Joel on Software , in the post Back to Basics : Shlemiel gets a job as a street painter, painting the dotted lines down the middle of the road. On the first day he takes a can of paint out to the road and finishes 300 yards of the road. “That’s pretty good!” says his boss, “you’re a fast worker!” and pays him a kopeck. The next day Shlemiel only gets 150 yards done. “Well, that’s not nearly as good as yesterday, but you’re still a fast worker. 150 yards is respectable,” and pays him a kopeck. The next day Shlemiel paints 30 yards of the road. “Only 30!” shouts his boss. “That’s unacceptable! On the first day you did ten times that much work! What’s going on?” “I can’t help it,” says Shlemiel. “Every day I get farther and farther away from the paint can!” Joel used it as an example of the kind of string processing naive programmers ar
2 0.83200639 266 andrew gelman stats-2010-09-09-The future of R
Introduction: Some thoughts from Christian , including this bit: We need to consider separately 1. R’s brilliant library 2. R’s not-so-brilliant language and/or interpreter. I don’t know that R’s library is so brilliant as all that–if necessary, I don’t think it would be hard to reprogram the important packages in a new language. I would say, though, that the problems with R are not just in the technical details of the language. I think the culture of R has some problems too. As I’ve written before, R functions used to be lean and mean, and now they’re full of exception-handling and calls to other packages. R functions are spaghetti-like messes of connections in which I keep expecting to run into syntax like “GOTO 120.” I learned about these problems a couple years ago when writing bayesglm(), which is a simple adaptation of glm(). But glm(), and its workhorse, glm.fit(), are a mess: They’re about 10 lines of functioning code, plus about 20 lines of necessary front-end, plus a cou
3 0.78395689 535 andrew gelman stats-2011-01-24-Bleg: Automatic Differentiation for Log Prob Gradients?
Introduction: We need help picking out an automatic differentiation package for Hamiltonian Monte Carlo sampling from the posterior of a generalized linear model with deep interactions. Specifically, we need to compute gradients for log probability functions with thousands of parameters that involve matrix (determinants, eigenvalues, inverses), stats (distributions), and math (log gamma) functions. Any suggestions? The Application: Hybrid Monte Carlo for Posteriors We’re getting serious about implementing posterior sampling using Hamiltonian Monte Carlo. HMC speeds up mixing by including gradient information to help guide the Metropolis proposals toward areas high probability. In practice, the algorithm requires a handful or of gradient calculations per sample, but there are many dimensions and the functions are hairy enough we don’t want to compute derivaties by hand. Auto Diff: Perhaps not What you Think It may not have been clear to readers of this blog that automatic diffe
4 0.77847105 2190 andrew gelman stats-2014-01-29-Stupid R Tricks: Random Scope
Introduction: Andrew and I have been discussing how we’re going to define functions in Stan for defining systems of differential equations; see our evolving ode design doc ; comments welcome, of course. About Scope I mentioned to Andrew I would prefer pure lexical, static scoping, as found in languages like C++ and Java. If you’re not familiar with the alternatives, there’s a nice overview in the Wikipedia article on scope . Let me call out a few passages that will help set the context. A fundamental distinction in scoping is what “context” means – whether name resolution depends on the location in the source code (lexical scope, static scope, which depends on the lexical context) or depends on the program state when the name is encountered (dynamic scope, which depends on the execution context or calling context). Lexical resolution can be determined at compile time, and is also known as early binding, while dynamic resolution can in general only be determined at run time, and thus
5 0.77731246 1716 andrew gelman stats-2013-02-09-iPython Notebook
Introduction: Burak Bayramli writes: I wanted to inform you on iPython Notebook technology – allowing markup, Python code to reside in one document. Someone ported one of your examples from ARM . iPynb file is actually a live document, can be downloaded and reran locally, hence change of code on document means change of images, results. Graphs (as well as text output) which are generated by the code, are placed inside the document automatically. No more referencing image files seperately. For now running notebooks locally require a notebook server, but that part can live “on the cloud” as part of an educational software. Viewers, such as nbviewer.ipython.org, do not even need that much, since all recent results of a notebook are embedded in the notebook itself. A lot of people are excited about this; Also out of nowhere, Alfred P. Sloan Foundation dropped a $1.15 million grant on the developers of ipython which provided some extra energy on the project. Cool. We’ll have to do that ex
6 0.7551555 272 andrew gelman stats-2010-09-13-Ross Ihaka to R: Drop Dead
7 0.75274932 597 andrew gelman stats-2011-03-02-RStudio – new cross-platform IDE for R
8 0.74491769 1807 andrew gelman stats-2013-04-17-Data problems, coding errors…what can be done?
9 0.7302869 1753 andrew gelman stats-2013-03-06-Stan 1.2.0 and RStan 1.2.0
10 0.71743298 818 andrew gelman stats-2011-07-23-Parallel JAGS RNGs
12 0.7115106 1134 andrew gelman stats-2012-01-21-Lessons learned from a recent R package submission
13 0.70829976 1799 andrew gelman stats-2013-04-12-Stan 1.3.0 and RStan 1.3.0 Ready for Action
14 0.70242095 1655 andrew gelman stats-2013-01-05-The statistics software signal
15 0.70080167 2161 andrew gelman stats-2014-01-07-My recent debugging experience
16 0.69896197 1036 andrew gelman stats-2011-11-30-Stan uses Nuts!
17 0.69799644 2020 andrew gelman stats-2013-09-12-Samplers for Big Science: emcee and BAT
18 0.69742227 1536 andrew gelman stats-2012-10-16-Using economics to reduce bike theft
19 0.68755889 867 andrew gelman stats-2011-08-23-The economics of the mac? A paradox of competition
20 0.68726844 1245 andrew gelman stats-2012-04-03-Redundancy and efficiency: In praise of Penn Station
topicId topicWeight
[(13, 0.012), (16, 0.083), (21, 0.027), (24, 0.185), (28, 0.024), (30, 0.014), (33, 0.011), (44, 0.053), (55, 0.025), (63, 0.02), (68, 0.017), (72, 0.026), (77, 0.028), (86, 0.033), (89, 0.091), (99, 0.23)]
simIndex simValue blogId blogTitle
1 0.9604851 1572 andrew gelman stats-2012-11-10-I don’t like this cartoon
Introduction: Some people pointed me to this : I am happy to see statistical theory and methods be a topic in popular culture, and of course I’m glad that, contra Feller , the Bayesian is presented as the hero this time, but . . . . I think the lower-left panel of the cartoon unfairly misrepresents frequentist statisticians. Frequentist statisticians recognize many statistical goals. Point estimates trade off bias and variance. Interval estimates have the goal of achieving nominal coverage and the goal of being informative. Tests have the goals of calibration and power. Frequentists know that no single principle applies in all settings, and this is a setting where this particular method is clearly inappropriate. All statisticians use prior information in their statistical analysis. Non-Bayesians express their prior information not through a probability distribution on parameters but rather through their choice of methods. I think this non-Bayesian attitude is too restrictive, but in
same-blog 2 0.9534086 2089 andrew gelman stats-2013-11-04-Shlemiel the Software Developer and Unknown Unknowns
Introduction: The Stan meeting today reminded me of Joel Spolsky’s recasting of the Yiddish joke about Shlemiel the Painter. Joel retold it on his blog, Joel on Software , in the post Back to Basics : Shlemiel gets a job as a street painter, painting the dotted lines down the middle of the road. On the first day he takes a can of paint out to the road and finishes 300 yards of the road. “That’s pretty good!” says his boss, “you’re a fast worker!” and pays him a kopeck. The next day Shlemiel only gets 150 yards done. “Well, that’s not nearly as good as yesterday, but you’re still a fast worker. 150 yards is respectable,” and pays him a kopeck. The next day Shlemiel paints 30 yards of the road. “Only 30!” shouts his boss. “That’s unacceptable! On the first day you did ten times that much work! What’s going on?” “I can’t help it,” says Shlemiel. “Every day I get farther and farther away from the paint can!” Joel used it as an example of the kind of string processing naive programmers ar
3 0.94659281 1991 andrew gelman stats-2013-08-21-BDA3 table of contents (also a new paper on visualization)
Introduction: In response to our recent posting of Amazon’s offer of Bayesian Data Analysis 3rd edition at 40% off, some people asked what was in this new edition, with more information beyond the beautiful cover image and the brief paragraph I’d posted earlier. Here’s the table of contents. The following sections have all-new material: 1.4 New introduction of BDA principles using a simple spell checking example 2.9 Weakly informative prior distributions 5.7 Weakly informative priors for hierarchical variance parameters 7.1-7.4 Predictive accuracy for model evaluation and comparison 10.6 Computing environments 11.4 Split R-hat 11.5 New measure of effective number of simulation draws 13.7 Variational inference 13.8 Expectation propagation 13.9 Other approximations 14.6 Regularization for regression models C.1 Getting started with R and Stan C.2 Fitting a hierarchical model in Stan C.4 Programming Hamiltonian Monte Carlo in R And the new chapters: 20 Basis function models 2
4 0.94603235 1702 andrew gelman stats-2013-02-01-Don’t let your standard errors drive your research agenda
Introduction: Alexis Le Nestour writes: How do you test for no effect? I attended a seminar where the person assumed that a non significant difference between groups implied an absence of effect. In that case, the researcher needed to show that two groups were similar before being hit by a shock conditional on some observable variables. The assumption was that the two groups were similar and that the shock was random. What would be the good way to set up a test in that case? I know you’ve been through that before (http://andrewgelman.com/2009/02/not_statistical/) and there are interesting comments but I wanted to have your opinion on that. My reply: I think you have to get quantitative here. How similar is similar? Don’t let your standard errors drive your research agenda. Or, to put it another way, what would you do if you had all the data? If your sample size were 1 zillion, then everything would statistically distinguishable from everything else. And then you’d have to think about w
Introduction: I remember in 4th grade or so, the teacher would give us a list of vocabulary words each week and we’d have to show we learned them by using each in a sentence. We quickly got bored and decided to do the assignment by writing a single sentence using all ten words. (Which the teacher hated, of course.) The above headline is in that spirit, combining blog posts rather than vocabulary words. But that only uses two of the entries. To really do the job, I’d need to throw in bivariate associations, ecological fallacies, high-dimensional feature selection, statistical significance, the suddenly unpopular name Hilary, snotty reviewers, the contagion of obesity, and milk-related spam. Or we could bring in some of the all-time favorites, such as Bayesians, economists, Finland, beautiful parents and their daughters, goofy graphics, red and blue states, essentialism in children’s reasoning, chess running, and zombies. Putting 8 of these in a single sentence (along with Glenn Hubbard
6 0.94256014 846 andrew gelman stats-2011-08-09-Default priors update?
7 0.93962079 1580 andrew gelman stats-2012-11-16-Stantastic!
9 0.93754482 693 andrew gelman stats-2011-05-04-Don’t any statisticians work for the IRS?
10 0.93750739 407 andrew gelman stats-2010-11-11-Data Visualization vs. Statistical Graphics
11 0.93554425 2086 andrew gelman stats-2013-11-03-How best to compare effects measured in two different time periods?
12 0.9347747 833 andrew gelman stats-2011-07-31-Untunable Metropolis
13 0.93414843 401 andrew gelman stats-2010-11-08-Silly old chi-square!
14 0.93315601 1473 andrew gelman stats-2012-08-28-Turing chess run update
15 0.92967635 1627 andrew gelman stats-2012-12-17-Stan and RStan 1.1.0
16 0.9294349 231 andrew gelman stats-2010-08-24-Yet another Bayesian job opportunity
17 0.92802632 1474 andrew gelman stats-2012-08-29-More on scaled-inverse Wishart and prior independence
18 0.92658889 2161 andrew gelman stats-2014-01-07-My recent debugging experience
19 0.92584634 807 andrew gelman stats-2011-07-17-Macro causality
20 0.92570275 788 andrew gelman stats-2011-07-06-Early stopping and penalized likelihood