Date:    2005-01-20
Author:  Robin Sabhnani
Purpose: examples to run usda project


-----------
data files
-----------
data.csv   : add case records so far from USDA
actual.csv : known pairs of similar or dissimilar cases

---------------
sample scripts
---------------
all.sh    : uses all features with predefined weights and feature types 
            from USDA data
subset.sh : uses subset of features from USDA data

----------------
example commands
----------------
# use variable method to define whether to run normalized 
# weight version or un-normalized 
#   method 1 = un-normalized
#   method 0 = normalized
./all.sh sim method 1
./subset.sh sim method 0

# by default for every query case, similarity measure with each other 
# case in the past is generated. use threshold ( don't print anything 
# less than threshold ) and top ( show only top n cases ) variable to 
# prune the total number of pairs generated.
./all.sh sim method 0 threshold 0.6 top 10

# use deg_1 and deg_2 variables to change cost slope for false alarms at ROC
# values in degrees ( note deg_1 > deg_2 ) defaults: 60, 30
./subset.sh sim method 1 threshold 0.5 deg_1 80 deg_2 10