Shifu 0.2.5 WoE Transform
In Shifu 0.2.5, woe and weightWoe are added in each column. And in ColumnStats, binWoe and binWeightWoe are also computed based on binning statistics. Missing value bin is involved in woe computing.
This feature is included in 'normalize' step. In ModelConfig.json#normalize part, two parameters are added: the first one is normType, by default is 'ZSCALE', it is the only normalizing method before Shifu 0.2.5, to set it to 'WOE', woe bin values will be used for normalization. The other parameter is 'isWeightNorm' means woe transforming on count woe bin or weight woe bin.
