Adatkelet

You could say I'm a sucker for bashing, but hell no, was I just curious.

Agile is something of an overused buzzword now and it seems to be the case now. Agile here equals with just installing a dozen frameworks after each other and skim another programming or markup language in a row spiced with some algos and stats. You gonna be exposed to SQL, NoSQL, Python, Avro, Pig, Flask, Bootstrap, Jinja, D3.js, nvd3.js, MongoDB, ElasticSearch, GitHub, dotCloud, S3, Elastic Mapreduce, Google Analytics, Mortar Data, ElasticSearch, Wonderdog, TF-IDF and Naive Bayes, all in 178 pages.

Thinking positively this is the quickest possible intro for a workflow with all typical phases, although weapons of choice seem arbitrary, never a why, never a pro or a con nor any alternatives. If you're lazy to understand the landscape and want a pre-cooked menu that you can cook, but can't change the recipes, go for it.

Check for yourself at O'Reilly.

Score: 2 of 5.

Mea culpa, but I was waiting for THE Pandas book given its author. I believed that pandas deserve a good intro book with decently built examples and learning curve, but I was wrong. This book is a very uneven kind of batched together webreference chapters. Most of the material is not meant for beginners sometimes even can get kind of scary. No pun intended, it smells like good intent, bad execution for me. As a starter Chapter 2 contained non-working code snippets while Chapter 3 exposed such inner workings that could confuse readers. Why is this here, I mean all editors went on strike? Truth to be told the 2nd edition fixed some of the most annoying bugs. Strictly appendix stuff is edited in as the main course. Handle with care — I will stick to pandas tutorials presented in IPython notebooks.

Check for yourself at O'Reilly.

Score: 3 of 5.

Faludi Bence előadása a Springfresh Django Meetup-on a CEU Microdata projekt Python-os ETL eszközétől, a mETL-ről (ejtsd: metil).

Január 23-án előadtam a Superweek konferencián, kösz Bánónak a hangért, nyilván a helyszínen merült le a hangfelvevőm.

A PyData szuperniche konferencia a javából a Python és az adatbindzsi metszetében. Tűéles előadások és tutorialok, alább a három kedvenc, de érdemes kotorászni a Vimeo csatornán.

Egy tanulság egy harminc perces videó helyett: NLP - don't do it. Do IR.

Nyéki Gábor tartotta a Data Science Meetup November-en a CEU Networks projekt első geek prezijét az adattisztításra épített irányított aciklikus gráf feldolgozónkról, a TARR-ról (Test and Rule Registry).

Interjúk a projekt vezetőivel, Szeidl Ádámmal és Koren Miklóssal.

Kicsit rövid a jelentkezési határidő - ma -, és a tízezer forintos adminisztrációs díj kicsit sok, de érdemes ráfeküdni. Részletek itt, válogatás a pályázat anyagaiból ott. (A rendezvényen rövid haknit követtem el, ami a hullámzó net miatt inkább bábozás volt - a végső és valódi Tableau-s Foursquare dashboarddal még mindig adós vagyok.)

I'm not a big fan of conferences although I've attended some, even presented at some in the international e-learning circuit. Strata, however always make me consider applying for a US visa - hint: Strata is just hitting London this October. Videos from Strata 2011 was already featured in this blog as they condensed such wit and knowledge with good presentation that I believe it's a must to consume them for any serious data nerd. This video product is not perfect - see my struggle with talking head videos before -, yes, in general it would make more sense to show the overhead projection in a big and the talking head in a small crop, but the Strata 2012 video compilation is actually more than you can handle and I'm not just talking about the 106 hours equaling gigabytes of MP4 movies. Yes, if you lurk around Youtube, you can find some of the best presentations out there, for free, but this product also contains the materials of the workshops that you definitely shouldn't miss. Please find below three of my favourite presentations from Avinash Kaushnik, Ben Goldacre and Jeremy Howard.

Check for yourself at O'Reilly.

Score: 5 of 5.

Ahogy erről az Index is beszámolt, a Youtube és a Facebook adatai alapján zenei slágerlistát készítettünk a 3G Multimédiával az Artisjus zenei jogvédő egyesület számára. A rangsort a Zene Ünnepe alkalmából a Dal+Szerző magazinban publikálták, ahol negyedévente frissülnek majd a toplista eredményei.