Usage
decTreeImpFunc(ITEM_SIZE, RANK_SIZE, CART_SIZE, DATA_FILE, INFO_FILE, OUTPUT_FILE, mode = c("E", "G", "S", "C"), TMODE = c("L", "P"), usePairwise = FALSE, isPW_DATA = FALSE, prediction = c("M", "F", "C"), NODE_SIZE = 400, MIN_NODE_SIZE = 400, RANK_ITEM = 2, ST_ALP = 0.95, CHI_ALP = 0.2, TRAIN_PROP = 1, useCV_TEST = FALSE, TEST_STAGE = 0)
Arguments
CART_SIZE
number of cross validation in tree pruning
DATA_FILE
please specify the input file as fullpath/input_file.txt
INFO_FILE
please specify the input file as fullpath/input_file.txt, note that the data should be
seperated by tab.
OUTPUT_FILE
name of the output file
mode
Spliting Criterion: E = Entropy G = Gini S = Statistical Test C = Chi-square test
TMODE
When mode = "C" P = Pearson Chi-square test; L = Likelihood ratio test
usePairwise
use pairwise comparison model or not
isPW_DATA
use pairwise data or top k-ranked data
prediction
M : mean, F : frequency, C : center
NODE_SIZE
usually one tenth of the number of observations
MIN_NODE_SIZE
min NODE_DIZE
RANK_ITEM
top-q measure (1-3)
ST_ALP
level of significance of Statistical test
CHI_ALP
level of significance of Chi-square test
TRAIN_PROP
proportion of training data; effective only when CV_TEST = false
useCV_TEST
use 10-fold CV testing or not
TEST_STAGE
the stage of the 10-fold CV testing; effective only when CV_TEST = true; value starts from 0 to V-1