A C++ implementation of Logistic Regression.
the book MLAPP(http://www.cs.ubc.ca/~murphyk/MLbook/) is recommended.
this program depends on liblbfgs(https://github.com/chokkan/liblbfgs).

compile:
g++ -O3 -o lr_train code/common_functions.cpp code/dummy_convertor.cpp code/logistic_regression.cpp code/train.cpp
g++ -O3 -o lr_predict code/common_functions.cpp code/dummy_convertor.cpp code/logistic_regression.cpp code/predict.cpp

comments:
1. -O3 is recommended when the dataset is large
2. if the g++ cannot find lbfgs lib, your can do it yourself: g++ -O3 -o lr_train code/common_functions.cpp code/dummy_convertor.cpp code/logistic_regression.cpp code/train.cpp /usr/local/lib/liblbfgs.a
3. the data formate: first column is the label(0/1), other columns are features(formate is feature_name:feature_value), please see the data/sample_train.tab and data/sample_test.tab
4. the conf/dummy.conf is configure for dummy variables index when map it to a array.