It’s actually a picture I took in 2013 during my Erasmus experience. More specifically, when I was doing a roundtrip to a music festival in Ísafjörður, Iceland.
Great video with Anthony Goldbloom (founder, and CEO of Kaggle) on the insights on statistical learning that Kaggle has been obtaining from its community (around 450 000 data scientists competing!).
The talk adreeses some techniques that seem to be tendencial winners in machine learning problems.
Hey! Finally, I am publishing here an online public version of my dissertation “Unwarping Heated Bones: A Quantitative Analysis of Heat-induced Skeletal Deformations Using 3D Geometric Morphometrics” for the fullfilment of Masters degree in Human Evolution and Biology at the Faculty of Sciences and Technology of the University of Coimbra. Hoping to get some commentaries now on it, so that I can incorporate really polished diamonds into the future scientific articles spawning from it.
First, you should notice that I have a repo on github where you can download all the materials to perform the complete data analysis that is published in my thesis manuscript. The reasons for letting others do this are 3-fold:
Giving back to the scientific community through sharing my data with all;
Validate my analysis through mass-peer-review;
Allow others to perform and test new ideas with my data or theirs (yo, but please say something if you are planning to do);
This does not only include the clusters of landmarks used for geometric morphometrics, but all the code in R to apply the methodologies used. For those who don’t know GitHub, it is the perfect environment for code-based scientific collaboration. It works through a system of pull and push where everyone can edit any info on the online folder, while maintaining a full version-controlled flow of work with an historic report of all the edits available and fully documented to the public. A true tool of Science 2.0.
Of course to make sense of it, most code is accompanied by commentaries. Yet, since information is never too much, my thesis is from today onwards out there as well. You just need to click here to read it.
Today was a very productive day. But something stood off, with a shiny feeling: a new tool for bioanthropologists is on the making.
That’s it folks. Osteomics is releasing a new app today, on one of the 4 basic pillars of Forensic Anthropology, one that we hadn’t explored yet: sex diagnosis. It is based on a famous article by Murail et al. (2005) that has probably the method with best balance between applicability and overall accuracy for sex estimation in osteology. Basically, we’ve implemented a new interface onto the same algorithm (that has been rewritten in R). Which is now, by the way, way faster than the original DSP Excell-based software.
It allows a case-by-case input, very useful if you are in the field, with your mobile phone or tablet, so you can access it and estimate sex right away. If you are doing more intensive laboratory work, it also has a population input mode through .csv tables upload, and it can calculate literally thousands of cases and generate an output of results in less than a second. Currently it is solely limited by internet access. This means, that contrary to DSP original version it is not Windows-exclusive, neither Excell-dependent!
Some print screens of DSP working in my iphone. The app allows you to do statistically significant sex estimations based on os coxae.
Since it is still on early stage (I will be adding a Manual with images tomorrow and other functionalities), I’d be very grateful if you guys tested it out, left any commentaries or try to crack it until you find some bugs for me to correct. After this starts to look as a final product, I will update it and add it to the Software section.
Hugs!
References:
Murail P, Bruzek J, Houët F, Cunha E. 2005. DSP: A tool for probabilistic sex diagnosis using worldwide variability in hip-bone measurements. Bulletins et mémoires de la Societé d’Anthropologie de Paris, 17, (3-4), 167-176.
Update!
22-October-2015 ~ around 01 AM
Manual with descriptions and pictures added;
Example dataset that works as a template for users to insert their data correctly, was also added;
Limited the number of minimum variables to 4 as in the original version of the software;
Corrected a lot of small bugs (calculations crashed under certains circunstances);
Added a lot of explanation texts here and there, and corrected some typos.
It is pretty much a final product by now. Any suggestions?
Today, I created a Zotero and Mendeley (.CSL 1.0.1) compatible reference style for the scientific magazines: Antropologia Portuguesa and Cadernos do GEEvH. My reason for doing this, is mainly because even though there was already an online version of this reference-style, it was only available for the paid-software EndNote. Since Zotero is free, it is definitely a more student-friendly option that I recommend to everyone. In the same spirit, I am sharing this for free to whoever might want or need it.
This .csl file uses the style described in Antropologia Portuguesa Guidelines. You can download it directly from here or by visiting the GitHub repo, an open collaborative coding platform where you can suggest modifications to the file.
Instructions:
Install Zotero if you haven’t done so.
Download the antropologia-portuguesa.csl from GitHub.
Double click on the file.
In Zotero’s settings, change your current style to “Antropologia Portuguesa”.