scottmsul

Couldn't someone just create a service that spoofs cookies in order to trick the airlines website into giving the lowest possible price?

Would it be possible to do the same thing in Haskell using partial function application, where the first argument is the name of the tag and the second argument makes the "<" ">" tags and applies string concatenation?

I wonder if cryptocurrencies could be used as an anti-spam prevention measure. In order to send you an email, maybe I have to send you 2 cents worth of cryptocurrency.

I was hoping to see some discussion about market-making bots or arbitrage bots, which I know for a fact can make money. Can these kinds of prediction bots actually beat random chance? I assumed price fluctuations mostly followed a random walk.

I've heard smart people say bitcoin will plummet because it has no intrinsic value. Well of course it doesn't! It's a currency. Currencies are an abstraction for value.

Sure, right now bitcoin is not a good day-to-day currency because of high transaction fees and slow verification speeds. But there are new altcoins that have solved both of these, that exist right now!

The real win for cryptocurrencies, in my opinion, is that it's public-key internet money, while debit cards are private-key internet money. With debit cards, you have to give someone your private key (16 digit combo+exp date+security code) in order to transact. Giving away your private key is prone to fraud, and the fraud-protection mechanism is to give a centralized authority access to everything you've ever purchased, and they look for anything "suspicious". On the other hand, with public-key money, when you give someone funds, there's no risk of them exposing you to fraud or unwanted transactions afterwards. Even though I understood the public-key mechanism for bitcoin early on, I didn't realize the full implications for the financial system until recently.

Would be interesting to see alpha zero trained on 2048, conv nets seem well suited for this kind of game.

For LIGO, mergers are just the beginning. LIGO isn't just a merger detector, but an entirely new window of astronomy. There's also teams looking for stochastic background signals from the big bang (similar to the CMB), or a constant hum from spinning neutron stars. And not to mention, if the sensitivity continues to improve, the moment of merging from BH-BH collisions could lead to new breakthroughs in GR.

This is a moonshot, but I would love to see a social network based on protocols similar to how emails work. Then different websites could implement interfaces for the protocol and talk to each other.

Company sizes follow a pareto distribution. In a pareto distribution, some small number x companies will be responsible for some large y% of anything.

This kind of simulation results in an exponential distribution, which is fairly equal compared all things considered. In an exponential distribution, most people have roughly the same order of magnitude of wealth (10s to 100s of dollars). In real life, the bottom 99% follow an exponential distribution pretty closely, while the 1% follow a pareto distribution, which is WAY More unequal. The transition is very sharp too, and has been studied in econophysics models.

Brief introduction to econophysics for the mathematically inclined: https://arxiv.org/abs/0709.3662

It think the edges might be cutoff by a few pixels.

A better idea would be to look for disagreements. Given a news article or claim, are there any sources out there which disagree? Then the user could browse both claims and decide for himself.

I think your link might be broken, it leads to a page that says "Sorry, but we can't find that page."

Location: Boulder/Denver CO

Remote: OK

Willing to relocate: Yes

Technologies: Data science, machine learning, Python, Haskell, SQL, Spark, Hadoop

Résumé/CV: http://www.drive.google.com/file/d/0B8XBeyOkf1TILUpHbUgtamRO...

Email: scottmsul@gmail.com

Github: http://www.github.com/scottmsul

LinkedIn: http://www.linkedin.com/in/scottmsul

-----------------------------------------------------

My name is Scott Sullivan, and I am a data scientist living in Boulder, CO. I recently completed the Galvanize Data Science Immersive in Denver, and before that I studied graduate physics at CU. For my Galvanize capstone project, I performed topic modeling of the news using matrix factorization. My final presentation is a live web app, feel free to explore it here: http://www.topicstory.com.

At Galvanize we performed several case studies, including:

  - Predicting the sale price of construction machinery
  - Predicting churn rates for a ride sharing company
  - Predicting fraud on an events website
  - Creating a recommender system for user-rated jokes

For each case study, we worked in small teams to implement a working solution in under a day. There was no guidance, requiring us to work as a team, assign tasks, share code, and combine our results in one presentation. Some example high-level tasks included cleaning data, performing EDA, interpreting features, and comparing different machine learning algorithms, including SVMs, trees, and neural nets.

Please let me know if you're interested!

Best, Scott

That's not exactly true, since bind and fmap have different types.

fmap :: (Functor f) => f a -> (a -> b) -> f b

bind :: (Monad m) => m a -> (a -> m b) -> m b

Functors simply map elements one-to-one, while monads map elements to new data structures, which have to be combined back to a single data structure.

However, it's good you mention the laws - it's certainly possible to define bind with the correct type but incorrect behavior. An example would be defining the List bind in a way that doesn't concatenate the elements in order.

The following is a really good reference for understanding not just the difference between functors, applicatives, and monads, but also why each is more powerful than the one before it.

https://en.wikibooks.org/wiki/Haskell/Applicative_functors#A...

We would need a way to transform a Tree of Trees into a single Tree. The inner Trees would be at leaf nodes of the outer Tree, so they could be attached in place.

Although I recall monads need to satisfy certain laws, and I don't recall if Trees satisfy them or not. Just because something is a data structure does not necessarily imply it is a monad.

Thanks, I changed the first sentence, hopefully it's less confusing.

It makes a List of Lists, then combines those into a List.

[5,3,4] becomes [f(5), f(3), f(4)], but f(x) must return a List. Those inner Lists are then concatenated.

When I say "original data structure", I mean the type of data structure is the same (in this case, List).

A monad is any data structure which implements bind. Bind is a higher-order function with two parameters - one is the data structure to be transformed, the other is a function which maps over elements in the data structure. However, unlike a normal map, each result of "bind" sits in its own version of the original data structure, which then have to be combined back into a single data structure. The way in which the data structures are combined is what makes each monad different.

For example, List is a monad. Suppose we had a List of Ints, such as [5,3,4]. If we were to run bind over this list, we would need a function that takes an Int and returns a List of something. We could use "show", the function which converts Ints to Strings (a String is technically a List of Char. Since this is a List, we're good). If we call bind using [5,3,4] and show, we get ["5","3","4"] which are then combined to "534".

We can check with the interpreter (>>= is bind):

Prelude> [5,3,4] >>= show

"534"

The equations from physics say "if we have some quantity X distributed among bins Y, and X is randomly exchanged between pairs of Y over and over, then P(bin y has amount x) is proportional to exp(-x)". In physics, X is energy and Y is atoms, while in economics, X is money and Y is people. If the money were truly being exchanged randomly, then the distribution should look exponential, that is, the probability person p has money m is proportional to exp(-m). What we observe is that the probability of having money m is actually m^(-alpha). Therefore the money is not being exchanged randomly, and the richest agents probably aren't lucky.

Right, power laws are extremely common, and all it says is that some aspect of the distribution is scale-invariant, but without saying why. For example, most companies are scalable, so it's no surprise that company valuations follow a power law. And since investors/founders own such large shares of companies, that would help explain why the 1% follows a power law. But as this article claims, if people gamble on log-scales, that would also create a power law. Power laws are so common and general, I find it difficult to believe that a power law is evidence of one way or the other.

I've read a bit into econophysics. In stat mech, the exchange of energy between atoms is assumed to be random. When the exchange of energy is random, the resulting distribution of energies is a Gibbs distribution, which means that the probability of atoms with higher energies falls off exponentially.

The application is that if you replace energy with money, and atoms with people, then the same equations hold for a model of people who randomly exchange money. Therefore if investments were truly random, you would expect the distribution of investors' wealth to follow an exponential distribution, not a power law! This directly contradicts the article.

Interestingly enough, the bottom 99% follows an exponential distribution, while the top 1% follows a power-law, and the transition is very sharp (eg, wealth plots have a "kink" in them).

Brief introduction to econophysics for the mathematically inclined: https://arxiv.org/abs/0709.3662

Scott Sullivan

Location: Boulder/Denver, CO

Remote: ok

Willing to relocate: Depends on the location

Technologies:

  - Languages: Java, C, MATLAB, Mathematica, Python, Ruby, OCaml, Lisp, Haskell, Octave
  - Software: Unix, Linux/Ubuntu, LaTeX, Microsoft Excel, Django, Postgresql
  - Skills: math, ML, data science

Resume/CV: https://drive.google.com/open?id=0B8XBeyOkf1TIZlhUZ29hMEtsa1...

Cover letter: https://docs.google.com/document/d/1D5SLszzsvR0R8iUoiTMzcyJe...

Email: scottmsul at gmail.com

Recently graduated from CU Boulder with a Masters in physics. Passionate about data science, machine learning, and web development. Highly motivated self-learner. Strong communication skills, especially with technical details - was a physics TA for several years, written technical papers for graduate school, given technical powerpoint talks.

Some examples of my work:

Predicting electric vehicle ownership / charging times given household power consumption: https://drive.google.com/open?id=0B8XBeyOkf1TIN1lLakludEtIV0....

Predicting the California Democratic primaries using Twitter data: https://drive.google.com/open?id=0B8XBeyOkf1TINFBkT1RhYkRqUV...

Scott Sullivan

Location: Boulder/Denver, CO

Remote: yes

Willing to relocate: I would prefer to stay in Boulder, but would relocate for the right company

Technologies:

  - Languages: Java, C, MATLAB, Mathematica, Python, Ruby, OCaml, Lisp, Haskell
  - Software: Unix, Linux/Ubuntu, LaTeX, Microsoft Excel, Django, Postgresql
  - Skills: math, ML, data science

Resume/CV: https://drive.google.com/open?id=0B8XBeyOkf1TIZlhUZ29hMEtsa1...

Email: scottmsul at gmail.com

Recently graduated from CU Boulder with a Masters in physics. Passionate about data science, machine learning, and web development. Highly motivated self-learner. Strong communication skills, especially with technical details - was a physics TA for several years, written technical papers for graduate school, given technical powerpoint talks.

An example of my work - predicting electric vehicle ownership / charging times given household power consumption: https://drive.google.com/open?id=0B8XBeyOkf1TIN1lLakludEtIV0...

One problem I have with this argument is that personal freedom isn't taken into account. Every time a decision is moved up one subsidiarity level, members at lower levels sacrifice a little bit of freedom.

For example, suppose the EU recognizes that gas-consuming vehicles contribute to climate change. From a utilitarian standpoint, it makes sense to ban all cars unless the economic advantage from driving outweighs the ecological damage from climate change. So perhaps the EU should ban all cars by default, but allow citizens to fill out a petition every time they need to drive. Except this is ridiculous, because we shouldn't need permission to drive. This is an assumed basic liberty (of course drivers should still pass a driver's test to get a license, but this only needs to be done once).

I think the utilitarian argument can still hold, but only if personal freedom is taken into account. It should be weighed like everything else, and recognized as a tradeoff. The United States allows freedom of speech, in exchange for giving people the right to post hateful comments online. The utilitarian might argue this is a bad thing, since society might be better off if hateful speech was removed from the internet. But in this case, I personally believe the benefits of freedom outweigh the drawbacks of offensive speech.

Every time a higher authority recognizes a prisoner's dilemma and enforces a rule, they are forbidding the players from playing the game themselves. Even if the societal benefits outweigh the drawbacks, there are hidden costs associated with sacrificing freedom.

Not only do you have match n numbers, you must also not match n+1, n+2, etc., otherwise you win a different prize.