def main(maxFeatures=30, maxDepth=8):
print "maxFeatures:", maxFeatures
print "maxDepth :", maxDepth
baseDir = globalConst.BASE_DIR
params = {
'max_depth': maxDepth,
'subsample': 0.5,
'verbose': 2,
'random_state': 0,
'min_samples_split': 20,
'min_samples_leaf': 20,
'max_features': maxFeatures,
'n_estimators': 500,
'learning_rate': 0.05
}
#'n_estimators': 12000, 'learning_rate': 0.002}
clf = GradientBoostingClassifier(**params)
# NOTE: first pass, no orderFile; 2nd pass, use orderfiles
test = Classify(trainFile=baseDir + 'workspace/trainMetrics.csv',
orderFile=useIfExists(baseDir + '/moby/corr32.csv'))
test.validate(clf=clf,
nFolds=2,
featureImportance=True,
outFile=baseDir + 'moby/trainPredictions.csv')
test.testAndOutput(clf=clf,
testFile=baseDir + 'workspace/testMetrics.csv',
orderFile=useIfExists(baseDir + '/moby/testCorr32.csv'),
outfile=baseDir +
'moby/testPredictions.sub') # NOTE .sub, not .csv
def main(maxFeatures=30, maxDepth=8):
print "maxFeatures:", maxFeatures
print "maxDepth :", maxDepth
baseDir = globalConst.BASE_DIR
params = {
"max_depth": maxDepth,
"subsample": 0.5,
"verbose": 2,
"random_state": 0,
"min_samples_split": 20,
"min_samples_leaf": 20,
"max_features": maxFeatures,
"n_estimators": 500,
"learning_rate": 0.05,
}
#'n_estimators': 12000, 'learning_rate': 0.002}
clf = GradientBoostingClassifier(**params)
# NOTE: first pass, no orderFile; 2nd pass, use orderfiles
test = Classify(
trainFile=baseDir + "workspace/trainMetrics.csv", orderFile=useIfExists(baseDir + "/moby/corr32.csv")
)
test.validate(clf=clf, nFolds=2, featureImportance=True, outFile=baseDir + "moby/trainPredictions.csv")
test.testAndOutput(
clf=clf,
testFile=baseDir + "workspace/testMetrics.csv",
orderFile=useIfExists(baseDir + "/moby/testCorr32.csv"),
outfile=baseDir + "moby/testPredictions.sub",
) # NOTE .sub, not .csv
def main():
baseDir = '/Users/nkridler/Desktop/whale/'
params = {'max_depth':8, 'subsample':0.5, 'verbose':2, 'random_state':0
'min_samples_split':20, 'min_samples_leaf':20, 'max_features':30,
'n_estimators': 500, 'learning_rate': 0.05}
#'n_estimators': 12000, 'learning_rate': 0.002}
clf = GradientBoostingClassifier(**params)
test = Classify(baseDir+'workspace/trainMetrics.csv')
test.validate(clf,nFolds=2,featureImportance=True)
