README_TREES

### Readme for the trees classifier in the gesture recognition application

The trees module implements the ensemble learning gradient boosting classifier based on decision trees.

### Configutation:
The following parameters must be set in config/personal.cfg to configure the classifier
estimators          - (int) number of estimators
learning-rate       - (float) learning rate
max-depth           - (int) max depth of decision tree
gestures            - comma separated list of gestures
queue_buffer_len        - length of recording buffer for life classification

The best results can be obtained by the following configuration:
estimators = 9
learning_rate = 0.2
max_depth = 2
gestures = 1,2,3,4,6
queue_buffer_len = 32

### Usage:
Go to folder src and start main application with 'python senseGesture.py'. Then use the following commands
u trees             - load tree classifier (if not specified by config file)
t                   - training of the gradient boosting classifier with specified config
v                   - validation of training, prints percentage of correct classified data
c                   - start live classification
e                   - exit application