betatim

There is however a kernel for C++ that you can try here: https://mybinder.org/v2/gh/QuantStack/xeus-cling/stable?file... if interpreted C++ is your thing :)

Once in a while someone working with/on Sage pops up on my radar but I am not sure what the up-to-date and maintained example is unfortunately.

https://jupyter.org/try all the repos linked from there are backed by mybinder.org (The interpreted C++ one blows my mind)

Some examples if you want to know how to configure things: * Install Python dependencies via requirements.txt https://github.com/binder-examples/requirements

* Jupyter with R and RStudio: https://github.com/binder-examples/r

* Julia support: https://github.com/binder-examples/demo-julia

* Installing additional APT packages: https://github.com/binder-examples/apt_install

(I work on Binder)

At the current level of usage and compute resources it costs about $50000 per year to run the services, if you don't have to pay the humans that help build and run it. We are an open project, if anyone wants to join to learn more about Kubernetes, Jupyter, Ops, Python, Docker you would be very welcome. We hang out on https://gitter.im/jupyterhub/binder and https://github.com/jupyterhub/binder is a meta repository.

(I am a project lead on Binder)

(I am a project lead on binder)

(I am a project lead on binder)

Tim is a scientist at the Large Hadron Collider and open source tool maker. I help develop open source software that is used by thousands of businesses and organisations to effectively model and learn about their data (http://github.com/betatim). I also lead a team of five to innovate on pattern reconstruction, particle detection, and experiment design which helped CERN develop a system able to process data at an unprecedented rate.

Tim consults on machine-learning, statistics and software development. Tim helps companies make data driven decisions by taking advantage of the data they have, building predictive models with that data, integrating expert knowledge, solving computational challenges and interpreting the results.

Blog: http://betatim.github.io GitHub: http://github.com/betatim

I consult on machine-learning, statistics and software development. I help companies make data driven decisions by taking advantage of the data they have, building predictive models with that data, integrating expert knowledge, solving computational challenges and interpreting the results.

Background: Tim has a PhD in Physics and several years experience as a post-doc working at CERN. My PhD research focussed on analysing large amounts of data using advanced statistical methods and machine-learning techniques. As a research fellow at CERN and EPFL I created and lead software teams responsible for designing the upgrade of the LHCb experiment. I value clear communication and often find myself interpreting between different groups of experts. I contribute to several open source projects which form the scientific python stack. I created a successful training program for scientists which dramatically reduces the on boarding time for new members of the collaboration.

Contact me for a free consultation.

Keywords: python, c++, scikit-learn, ROOT, jupyter

Web: https://betatim.github.io Twitter: https://twitter.com/betatim

The author wrote a second column (http://www.nytimes.com/2014/07/27/your-money/heads-or-tails-...) explaining how they came up with the number of funds that would outperform the market five years in a row.

The reasoning goes like this: 2/2862 funds made it (or about 0.07% of funds).

The author then says: If you assume a fund has a 50% chance of beating the market, and a 50% chance of falling behind then the chances of one beating the market five years in a row would be 0.5^(5 * 2). They just state that they flip a coin twice per year per fund (hence 0.5^10), can someone explain why?

This is a link to nature's new "share it for everyone to read if you have paid access to the article" version of 'open-access'.

A new antibiotic kills pathogens without detectable resistance, Nature, January, 2015. DOI: 10.1038/nature14098

If you want to learn about graphene's amazing properties go ahead and read the paper here: http://rdcu.be/bKud

For example for http://www.nature.com/nature/journal/vaop/ncurrent/full/natu... the readcube link is simply http://www.readcube.com/articles/10.1038/nature14015

It seems too easy? What am I missing?

In the case that they differ you almost always find that you have very few observations. I would argue that this 'difference' is not that exciting because it must be dominated by your assumptions, not your observations. After all once you accumulate enough observations the two methods tend to converge.

Personal conclusion: if the methods disagree work on getting more data instead of fighting over which method is better.

Struggling to come up with a simple example. Imagine you collect some data, analyse it and see a "3sigma" effect. You decide to collect more data to see if the effect keeps getting bigger or goes away. After collecting a lot more data you get "5sigma". This could be because it is real, or because you have an "unknown unknown"

If you had a systematic (as oppose to a random effect) shift in your analysis, collecting more data will make you more sure that there is an effect. Even though all you are seeing is the effect of the systematic shift.

For almost all analyses in particle physics (or astro or ...) we spend a huge part of our time evaluating "systematics", it isn't uncommon that this part of the analysis takes a lot longer than the nominal result. Unfortunately this only protects you against known unknowns. You can't take into account things you don't think of/check for.