Re-design for sklearn

[juifa-tsai]

The format of module is not completely compatible to sklearn. The reason for adopting to sklearn is because of the parameter optimization. Sklearn provides well-defined tool for tuning parameter and validation. Thus, making the module to adopt sklearn is kind of way to simplify future work.

1. Change get_params() to output the default parameter sets with {'name':value} instead.
2. Add relevant function to output results.
3. Add set_params() to any model for implementation grid search in the future, instead of update each parameter.
4. Add score() to every models.