Big Data Hackathon Marathon after game summary

Last weekend participated in the IBM Analytics held "Big Data Hackathon" Beijing Station competition, 4 people team to get the first place, very happy, also very difficult, we four wesor like to work together for a long time, their respective cooperation with tacit understanding, Won the final trophy. The first day from 9 o'clock to 11 o'clock more, I would like to stay up late but limited to the organizer of the venue, the next day 9 o'clock to three o'clock in the afternoon, even eat a plate against the computer. These two days tired has been relieved but come, sleep all feel tired of sleep, but did learn a lot of things, from the game, from the teammates ...

As the doctor said, it is important to persuade others to feel that their work. The remark had been seen before from elsewhere, and it was now more reasonable to think otherwise. The work we do is big or small, how to tell others, our work is very important, actually is to say, we have to solve the problem is how important and difficult to put their results described very useful. And our work is always a process of convergence. The problem described above is grand, and then it needs to be convergent in one step, then the problem is summed up to the point to be solved, the key problem is solved and the whole problem is solved.

The solution to the problem is not how complex, with the simplest algorithm to solve the most difficult problem, is the most commendable. So don't always think about what the new algorithm, the classical algorithm to clear the application, solve the actual problem.

Our PPT preparation is good, also not white write a few months of PPT to various leaders report, the story is round, and no nonsense, praise. I think of myself to go to a country to do thesis speech, for several years not to speak, after the opportunity also need to come on stage to talk about.

Time is tight, Scala code can only be realized function, there are a lot of places to optimize. And just use the spark core parts, in fact, can be used in mllib some advanced data type Dataframe to preprocess data, unfortunately has not learned, Collaborativefilter is now learning to sell, need to speed up spark core + Mllib's learning progress.

GitHub has learned to use, indeed very well.

Machine Learning Algorithm learning can not stop, fortunately this is no problem data limited competition, if there is a chance to participate in the competition, the machine learning model algorithm requirements are high, while it is a little bit of foundation, learn more practice, do not want to know the mathematical basis of each algorithm, But to know what the problem with what models and algorithms, how to optimize the adjustment and so on.

A few wosor really "Sao" taste congenial, even if the last not to take the prize, two days is also particularly happy, delicious good drink, even squat pits are together, glad that they have these good friends, proud of themselves is also a part of the son.

are working on new data, translating new model algorithms, learning new things, starting with new goals ...

http://www.csdn.net/article/a/2015-08-18/15827301

Http://www.c114.net/news/212/a914245.html

END

Big Data Hackathon Marathon after game summary