Kaggle is the place for Data Scientists. If you want to learn about Machine Learning, Data Mining and data hacking you should definitely visit Kaggle.
We decided to participate in the ongoing competition: Springleaf Marketing Response. Eventually it improved our feature enginerring, Data Mining and the FX trading. Facing large data sets is very common in Kaggle, on the other hand, in the FX market we have got a lot of data so there is a lot to learn from Kaggle regarding the FX market.
One of the most important aspects of Data Science is Feature Engineering: the art of selecting, transforming and messing around with our features. In Kaggle the training sets are super wide and super long. The curse of dimensionality is unavoidable here. We must select a feature subset which will be the best representation of each and every instance.
Kagglers tend to incorporate several tools which create a Victorinox. First it is very important to visualize the data and perfectly know what is the temperament of your data set. Then we proceed with removal of outliers or non descriptive, biased and ambiguous features. Finally we have to discretize or hash the non numeric values, because most of the cool classifiers tend to prefer numerical data.
When dealing with large data sets, Python or R are the way to go for quick and real-time solutions. Kaggle allows you easily play with the data, make submissions and use the most known libraries for Machine Learning, from your browser, anywhere, anytime and instantly.
Two very interesting and helpful sites that come along with Kaggle are dataquest and DataCamp. Those websites provide free introduction courses in Python and R programming on the fly. Also they have special relation and examples regarding Kaggle.
Finally the data is out there and the tools are out there, so it's time to explore! Check out our performance in Kaggle. Learn more about Scientific FX Trading.
