Ask HN: Learning multiple languages vs. mastering one | Hacker News
Scala, Clojure, even Haskell or Go look like they might be fun, but clearly, "next" for someone doing "stats and viz" with R is (1) Python, (2) JavaScript, (3) more bash, and perhaps (4) advanced SQL / DB languages.
programming  data:science 
13 hours ago
If we add terms (e.g. interactions, polynomials etc.) to impove model fit suggested by residuals plots, is this classed as data dregding? - Cross Validated
"… if you can't assume the model is simple up front, and you entertain the notion that it's not simple, you have to either (1) just allow it to be flexible and stick with that flexible model, (2) be Bayesian, or (3) be aggressive but bootstrap yourself to adjust for the aggression." -- Frank Harrell.
stats:models  stats:bayesian 
13 hours ago
LSE Thinks | If previously in force, would GDPR have prevented the use of Facebook data? - YouTube
The key GDPR thing *is* the possibility of class action-style litigating. Completely missed by e.g. Mozilla [1], even though any real change from GDPR is more likely to come from human-rights groups -- e.g. [2,3] -- than from individuals.

[1] https://blog.mozilla.org/internetcitizen/2018/05/23/gdpr-mozilla/
[2] https://www.ldh-france.org/protection-donnees-personnelles-double-avancee-rgpd-convention-108/
[3] https://noyb.eu/

@seealso https://www.youtube.com/watch?v=Zt5H0t_KXOA
web:social-networks  web:privacy 
14 hours ago
Context Compatibility in Data Analysis · Simply Statistics
"Data linkage is a common place to encounter context mismatches because rarely are different datasets collected with the other datasets in mind. Therefore, careful attention must be paid to the contexts within which each dataset was collected and what assumptions or modeling most be done in order to achieve context compatibility."
stats:spatial  data:science 
2 days ago
Modern modelling techniques are data hungry: a simulation study for predicting dichotomous endpoints | BMC Medical Research Methodology | Full Text

… We compared three relatively modern modelling techniques: support vector machines (SVM), neural nets (NN), and random forests (RF) and two classical techniques: logistic regression (LR) and classification and regression trees (CART). We created three large artificial databases with 20 fold, 10 fold and 6 fold replication of subjects, where we generated dichotomous outcomes according to different underlying models. We applied each modelling technique to increasingly larger development parts (100 repetitions). The area under the ROC-curve (AUC) indicated the performance of each model in the development part and in an independent validation part. Data hungriness was defined by plateauing of AUC and small optimism (difference between the mean apparent AUC and the mean validated AUC <0.01).


We found that a stable AUC was reached by LR at approximately 20 to 50 events per variable, followed by CART, SVM, NN and RF models. Optimism decreased with increasing sample sizes and the same ranking of techniques. The RF, SVM and NN models showed instability and a high optimism even with >200 events per variable."

@via http://www.fharrell.com/post/stat-ml/
stats:models  stats:machine-learning 
2 days ago
RmS < Main < Vanderbilt Biostatistics Wiki
> Regression Modeling Strategies
> With Applications to Linear Models, Logistic and Ordinal Regression, and Survival Analysis

Frank Harrell.
r  stats:regression 
2 days ago
Navigating Statistical Modeling and Machine Learning | Statistical Thinking
"1. Do you want to isolate the effect of special variables or have an interpretable model? If yes, turn left toward SM [stat. modeling]; if no, keep driving …
2. Is your sample size less than huge? If yes, park in the space designated “SM”; if no, …
3. Is your signal:noise low? If yes, take the ramp toward “SM”; if no, …
4. Is there interest in estimating the uncertainty in forecasts? If yes, merge into SM lane; if no, …
5. Is non-additivity/complexity expected to be strong? If yes, gun the pedal toward ML; if no, … you can continue the journey with SM."

@seealso http://www.fharrell.com/post/stat-ml/
stats:models  stats:machine-learning 
2 days ago
EUROPP – The death of ‘business as usual’ in the EU
"Ironically, it is precisely these member states whose publics are more Eurosceptic that are also more successful in weakening, or watering down, these CSRs during the amendment phase in the Council. And it is also these member states which are least likely to actually act on the Commission’s recommendations. While this raises questions about how the Commission responds to such activities in the Council and vice versa, it also sheds light on just how deeply the EU has been penetrated by the forces of politicisation: even in relatively technocratic procedures like the European Semester, politicisation affects the behaviour of individual actors and the interaction between them."
eu:politics  eu:economy 
2 days ago
How to Survive a Laptop Fire
"Everything on my local hard-disk is disposable." -- I get the idea, but one consequence is that everything you own is "double-owned" -- some corporation also owns your stuff online. If that corporation goes dark, you lose your stuff; if that corporation backdoors your stuff, you share it forcibly.

Is that problematic? Yes, in both scenarios, because it damages your sense of ownership and your sense of privacy, two things that we actually need to function properly. Without ownership, we become addicted to the networks that replace possession -- and that's definitely not as desirable as a sense of private property, which is, to some extent, a component of privacy. (Not everything about private property is right, but some of it is.)

The loss of privacy is largely a side-effect of the point above, and the combination of both is terrible -- the mental equation is just toxic, unlike sharing some things voluntarily (see creation of public goods like F/LOSS, commons etc.).
mac:backup  web:social-networks 
3 days ago
Facebook management moves around but people don’t leave - Recode
"It’s sometimes hard to grasp Facebook’s massive scale. The site has more than 2.2 billion users worldwide. In fact, there are nearly as many people who actively use Facebook every month as there are followers of Christianity. Facebook’s 2017 revenue — around $40 billion — was more than the GDP of about 100 different countries. It’s not an exaggeration to think that second- and third-tier Facebook executives have a chance to impact more lives than most of the world’s elected politicians."
3 days ago
Does "status threat" explain the 2016 presidential vote? - Statistical Modeling, Causal Inference, and Social Science
Elementary points on how to read regression coefficients.

Also, this:

"The z-score, or p-value, or statistical significance of a coefficient is a noisy random variable. It’s just wrong to equate statistical significance with reality. It’s wrong in that the posited story can be wrong and you can still get statistical significance (through a combination of model misspecification and noise), and in that the posited story can be right and you can still not get statistical significance (again, model misspecification and noise). I know this is how lots of people do social science, but it’s not how statistics works."
polisci:elections  usa:politics  stats:regression 
3 days ago
FreshRSS, a free, self-hostable aggregator
Recommends PHP7, though PHP5 should work.
4 days ago
Banlieues : deux députés relancent le débat sur les statistiques ethniques
"Chargés d’interroger l’action de l’Etat dans l’exercice de ses missions régaliennes dans le « 9-3 », les auteurs confient avoir été confrontés, « dès les premières auditions », à « une méconnaissance de la population », à commencer par le nombre réel d’habitants du département francilien."
fr:ethnic-politics  soc:quantification  fr:immigration 
4 days ago
azat-co/fullstack-javascript: Source code for the Fullstack JavaScript book
"Full Stack JavaScript: Learn Backbone.js, Node.js and MongoDB, 2nd and 3rd Editions"
4 days ago
[1805.06005] Reconstructing mesoscale network structures
"When facing complex mesoscale network structures, it is generally believed that (null) models encoding the modular organization of nodes must be employed. The present paper focuses on two block structures that characterize the mesoscale organization of many real-world networks, i.e. the bow-tie and the core-periphery ones. Our analysis shows that constraining the network degree sequence is often enough to reproduce such structures, as confirmed by model selection criteria as AIC or BIC. As a byproduct, our paper enriches the toolbox for the analysis of bipartite networks - still far from being complete. The aforementioned structures, in fact, partition the networks into asymmetric blocks characterized by binary, directed connections, thus calling for the extension of a recently-proposed method to randomize undirected, bipartite networks to the directed case."
networks:bipartite  networks:null-models 
6 days ago
Yiqing Xu | Software - interflex: Producing Flexible Marginal Effect Estimates with Multiplicative Interaction Models
"This package performs diagnostics to assess assumptions of multiplicative interaction models (namely, linear interaction effect and common support) and impliments [sic] flexible estimation strategies that allow for nonlinear interaction effects and safeguard against excessive extrapolation."

@reference http://yiqingxu.org/papers/english/2018_HMX_interaction/main.pdf
r  stats:stata  stats:regression  academia:publishing 
6 days ago
Economic development and democracy: An electoral connection - KNUTSEN - - European Journal of Political Research - Wiley Online Library
"… only election‐centred indicators are robustly associated with economic development. […] Further analysis shows that development affects electoral democracy by reducing electoral fraud, election violence and vote buying."
polisci:democracy  polisci:elections 
9 days ago
Why combatants fight: the Irish Republican Army and the Bosnian Serb Army compared | SpringerLink
"The study demonstrates that nationalism played a relatively marginal role in combatants’ motivation to fight. Instead our research indicates that individualist motivations, small group solidarity, and local networks dominate."
violence:war  polisci:ideology 
9 days ago
Attack When the World Is Not Watching? US News and the Israeli-Palestinian Conflict | Journal of Political Economy: Ahead of Print
"We find that Israeli attacks are more likely to occur when US news on the following day is dominated by important predictable events. Strategic timing applies to attacks that bear risk of civilian casualties and are not too costly to postpone. Content analysis suggests that Israel’s strategy aims at minimizing next-day coverage, which is especially charged with negative emotional content. Palestinian attacks do not appear to be timed to US news."
world:israel-palestine  world:israel 
9 days ago
Reproducibility of research: Issues and proposed remedies | PNAS
" The Colloquium was organized by David Allison, Richard Shiffrin, Victoria Stodden, and Stephen Fienberg [†]."
academia:research  via:cshalizi 
11 days ago
The Matthew effect in science funding | PNAS
"… Our results show that winners just above the funding threshold accumulate more than twice as much funding during the subsequent eight years as nonwinners with near-identical review scores that fall just below the threshold. This effect is partly caused by nonwinners ceasing to compete for other funding opportunities, revealing a “participation” mechanism driving the Matthew effect." -- I have seen that mechanism at work in France: young researchers spend huge energy and time on the first project, get rejected, depressed and black-sheeped by the top brass, and never submit again.
academia:funding  via:cshalizi 
11 days ago
Mass Mobilization in Autocracies Database
"The Mass Mobilization in Autocracies Database (MMAD) contains sub-national data on mass mobilization events in autocracies worldwide."
data:social-science  polisci 
11 days ago
Comment les « leaks » ont changé la façon de travailler de la justice
"Bercy, qui a déclenché au total 411 contrôles fiscaux dans cette affaire, préfère généralement traiter les cas de fraude fiscale par des redressements avec pénalités financières à la clé, sans les transmettre à l’autorité judiciaire. Un choix qui leur appartient, et qu’on surnomme le « verrou de Bercy ». Mais qui n’a pas pu jouer en l’espèce, les informations étant déjà dans l’espace public."
fr:economy  taxation  corruption 
11 days ago
Nous avons rencontré Josep Borrell – Le grand continent
"La social-démocratie a perdu le contact avec les classes populaires qui lui garantissaient de compter au moins pour 40% de son électorat. Auparavant, les sociaux-démocrates étaient autour de 40% dans toutes les élections en Europe. Désormais, ils n’obtiennent plus que 20%, au maximum – et beaucoup moins en France et aux Pays-Bas. Sir Julian Pristley, qui fut secrétaire du Parlement européen, avait écrit un article intitulé « Getting Used to 25 percent » en 2016. C’était un avertissement pour les sociaux-démocrates."

-- With much more stuff (Catalonia, Eastern European countries, …).
eu:politics  polisci:elections  polisci:ideology  world:spain 
13 days ago
« earlier      
academia academia:homepages academia:humour academia:phd academia:profession academia:publishing academia:research academia:teaching anthropology art art:comics art:film art:illustration at_a_loss_for_tags books cancer copyright:contenders copyright:control copyright:open-access css data:availability data:economics data:health data:journalism data:json data:management data:mining data:science data:social-science data:sql demography design econ econ:history econ:macro econ:theory epidemiology epistemology epistemology:causality eu:economy eu:politics fr:copyright-control fr:economy fr:health-policy fr:higher-ed fr:history fr:politics fr:society games geography germany:politics github google history history:ussr html html5 immigration imported inequality inequality:gender inequality:income javascript journalism language literature lulz lulz:journalism mac:osx mac:software mac:terminal mac:text-editing mac:utilities markdown mathematics medicine music music:hard-rock music:hip-hop music:techno networks paris pathetic pdf philosophy photography polisci polisci:democracy polisci:elections polisci:ideology polisci:ir polisci:parliaments polisci:parties polisci:polling-institutes polisci:power polisci:protest polisci:state political-economy programming psychology public-health public-health:tobacco-control python r religion science soc soc:collective-action soc:health soc:quantification soc:theory social-science stats stats:bayesian stats:cluster-analysis stats:correlation stats:courses stats:intermediate stats:introduction stats:logit stats:machine-learning stats:measurement stats:models stats:multilevel stats:neural-networks stats:pca stats:probability stats:regression stats:spatial stats:stata stats:surveys stats:survival stats:teaching stats:text-analysis stats:theory stats:time-series taxation tech:security tex tex:bibliography tex:writing text:manipulation text:mining typography uk:health-policy uk:higher-ed uk:politics uk:scotland uk:society usa:congress usa:economy usa:empire usa:health-policy usa:higher-ed usa:politics usa:presidency usa:society usa:surveillance usability via:cshalizi via:henryfarrell via:kjhealy via:sechilds via:waxy via:xmarquez videogames violence:crime violence:drugs violence:prison violence:racism violence:terrorism violence:war viz viz:d3.js viz:design viz:diagrams viz:distributions viz:ggplot2 viz:maps viz:networks viz:shiny viz:tables viz:timelines web:architecture web:blogging web:browsers web:facebook web:networks web:social-networks web:twitter web:wikipedia webdesign world:algeria world:china world:iraq world:israel-palestine world:libya world:morocco world:politics world:russia world:syria writing wtf

Copy this bookmark: