I Uploaded my First AUR Packages

I have been using the Arch Linux ever since it was recommended to me in Germany by somebody at the language meetup. I am kind of embarrassed, that I do remember him, but I have no idea what his name is.

AUR has become a place where I have been searching for programs not the the repositories. I guess like most of the people that use Arch. So far, most of the programs that I wanted were there, except a couple of python libraries.

Well, there is always pip for libraries, but doing some through pacman and some through pip seems to me like a recipe for a mess. And possible conflicts.

So, if possible, I have always tried to create the PKGBUILD file and installing it from there. But I never really put it in the AUR, mostly from fear.

Well, now I did. You can check it on the AUR page for python-stackapi-git and python2-pytumbl-git.

The MBTI Analysis of the StackOverflow

So now I think I am coming close to the version of MBTI analysis of the StackOverflow that I am going to be presenting. There are some potentially interesting things, that I might take the time to write upon them at the later time. If I fine the time. For example, I am really interested why the INTJ are so interested in the R language compared to the other types.

You can check the code and the results here. Or you hope that I will write about them at the later time.

Splitting the StackOverflow Data

For our Data Mining project, me and my teammate were analysing the StackOverflow data on Kaggle. But since she was dead set on not installing anything on the computer, we needed to find a different solution. Especially since the Data Science workbench that she was using had a stingy file limit.

I know there is a reason for that limit, as they don't want free customers abusing the service. I just find it impractical, so I am not using it.

Either way, we needed to find another way, so I created a script that could filter the data based on the tags and create the new files. This way she would at least make language specific analysis. The code for this can be found here

Using StackOverflow API

The teammate that I got for my Data Mining course was not that interested in the data analysis that I was doing, so I asked her what she wanted to do. She said that she wants to analyse the StackOverflow data.

So in order for us to do this, I tried to find an API that we could use. I ended up doing a small example of how to use the API, to get her started. The code can be found here.

We did not end up using the API, but it was still nice found out that it is possible.

Incrementalism or Big Changes: Which One is Better to Change Lifestyle

I sometimes really wonder which one of these two is a better way. On one had, the incrementalism has the pros that it does not make me experience the difference in my everyday life. On the other hand, the incrementalism is not really something that I could stick with on the long run.

On the other side, making big changes to the lifestyle usually does not make it stick either, as the force has to be big enough to overpower all the inertia and the resistance. In most cases, it is hard to summon this kind of power.

I actually like to sort of combine them. I am using the big changes, in order to bring the incrementalism to my life. Every time that I do some big change, I know that I am most likely not being able to stick with it. But each time I do a change like that it will bring the change in that direction to my everyday lifestyle.

One of the example was my experiment with not having the access to the internet until 19 o'clock. I have does this experiment for, if I remember correctly, 2 months. I still remember that before, one of the first things that I would do in the morning is starting a computer and check my email.

These day, if I actually have work to do, it is not hard to spend time without the internet for the first couple of hours. Also, before, if I thought I would be without internet for just a couple of hours, I would try to make sure I had enough things to do. These day, I know that it is very unlikely that I run out of work to do, even if I try to keep the minimal amount of work on my computer.

The other one that I used it was food. If I am comparing what I am eating now with what I was eating years ago, I am eating a lot less processed food. But most of my eating experiments ended up with my going back to my normal eating patterns (one exception was giving up meat, not including fish).

The third one was my social life. I would try to do these big changes from time to time, and sometimes along these lines, I had the social life that was quite abounded for me. It was a little easier, because I don't need a lot of socialization, and I don't want too many socialization. But I still want some, and at some point, I actually gained some social skills.

This could be used to many other things. Becoming more productive, increasing exercising time and many others. Right now I am trying to use it for learning. We will see how well it goes.

Changing Habits Based on Different Paradigms

When I was doing the seminar paper on the nudging for Actors, Behaviours and Decision Processes class, I made the list of all the ways that people can get influenced. So when I was looking through my notes, I noticed that these ways can actually helps us find different methods in order to change our habits.

I will try to explain this on the example of a person trying to start exercising.

The first one is based on the economic incentives. So there would be some financial incentive for exercising. For example, every day that a person does not go to the gym, they have to pay 10 euros to a friend. Or every day that a person does not go for a run, they donate 10 euros to a political party they disagree with.

The second one is based on the social pressure. This one would have a person change their social circle in a way, that there would be more people exercising regularly in it. Some of the ways to do this is to join meetup for people that exercise, like the cycling group. The other one is to take up a class where the whole group exercise at the same time. Or become pat of the team in a team sport, like the volleyball, basketball or football.

The third one is based on the psychological biases. This one would, for example, use different biases that we have to make it more likely to exercise. Some of these things are like wording the question differently, like instead of "Are you going to exercise today?", the question could be "Are you going to a run or cycling today?". Different apps that use gamification for the exercise can also work. For example, the app that counts the amount of exercise each day and they compares it with previous days or other people.

The fourth one is changing the environment, sort of like using the embodied paradigm or situated cognition in the cognitive science. Make sure there is the running gear out and ready each morning. Move to a place where people are constantly exercises (or change the job to get into that environment, but that seems a bit extreme :) ). Or simply change the circumstances, like sell you care and force yourself to take the bike to work. Which works even better if there is not good public transportation around.

Sure, it does also helps if a person can find something enjoyable to do. For example, I can't stand running or going to the gym. But I like walking, cycling and swimming. But once a person finds something like that, it is helpful to use at least some of these methods.

For example, for exercising, I know that if I don't the monthly/semester ticket, then I am going to cycle everywhere. And these short distance do add up to about 100km per week, which I think it quite a nice number. Here I am using the mixture of third (default choice) and forth point, by making the bike the most obvious choice. Also I experience some of the effects of the first point, as I have more money left over, as I did not spend it on the ticket.

Which one are you using to change your habits? If you so far relayed only on the will-power, than try some of these options.

Dealing with Unbalanced Data - Averaging Results with MBTI Data

With my MBTI project, I have do deal with hugely unbalanced data. So I have researched how I can deal with it. I have already used the scaling techniques, where I simply used the same amount of units in each class.

This time I decided to try extending this. I created multiple models, each time with the subset of data that was balanced. After that I averaged it.

You can find the code here.

For Better Thinking, Think in a Foreign Language

For one thing, we know that we are not really rational homo economicus that economics like to assume we are. Be it because we don't have resources to be like that, leading to bounded rationality, or because it is much more energy efficient to have two system, with only one being rational.

This can lead to a lot of biases. From sunk cost bias, where people prefer to stick with things that they invested a lot in, or status-quo bias, where people prefer the known.

And while these biases can sometimes be helpful, sometimes we want to be able to make decision without them influencing us. And there is one pretty simple way to do it: simply think in a language that is not your native tongue.

A couple of examples were written in the this 2012 paper. They showed that presenting things in a foreign language can diminish the biases compared to the native language.

They showed this works with framing bias, where people make different decisions, based on how the things are worded, and loss aversion, where people prefer to not lose than to gain.

They claim that this might be the effect of the emotional distance created by using the less fluent language. Which might be true, at least when we take into account this 2014 article.

In that paper they were comparing the more moral decision making. People were making more utilitarian decisions making. In the morality, it is said that instinctive or system 1 morality is the one that protects someone's rights. The rational or system 2 morality is the one that is working for the greater good.

Since utilitarianism is based on the working for the grater good, it is considered less emotional and more rational.

For for the people not believing that there can be differences based only on the language use: Kobayashi at al. showed that there are differences in brain activity, when people were doing the exercise in the native and foreign language.

So next time you try to make a more rational decision, maybe give it a try and do it in a foreign language.

Getting MBTI Tagged Text from Tumblr

I decided that I am putting some of my code that I used last semester in the jupyter books, because I think that this makes it more readable. So I decided to refractor the code that I used and put it here.

I only tested on the small samples, so I might have missed something. But I think it should work.

You can find the code here.