data mining issues are the performance, accuracy, robustness, scalability and interpretability of the minor and the relations between them. i.e. you must be careful not to increase the performance and on the other hand lose the accuracy of the minor.

for the data repository i suppose you to visit this site

http://archive.ics.uci.edu/ml/datasets.html

I hope my answer helped you.

I suggest you to read this paper:

[1] Bottou, Léon, and Olivier Bousquet. "The Tradeoffs of Large-Scale Learning." Optimization for Machine Learning (2011): 351.

This pre-print is also nice:

http://arxiv.org/abs/1308.1479