os.chdir("..") # set to parent dir
# ===============================================================
# Global params
# ===============================================================
MAX_ROUND = 1200
# ===============================================================
# Load and downcast
# ===============================================================
train, x_test, sub = load_all()
x_train = train.drop('target', axis=1)
y_train = train.target
x_train, x_test = downcast(x_train, x_test)
del train
gc.collect()
# ===============================================================
# Run cv and bags
# ===============================================================
n_splits = 5
folds = list(KFold(n_splits=n_splits, shuffle=True, random_state=47).split(x_train, y_train))
trn_gini, y_trn, y_tst, fscore = etc07(x_train, y_train, x_test, folds, MAX_ROUND, n_splits)
# ===============================================================
# Save submission and figure
# ===============================================================
if MAX_ROUND > 300:
sub.target = y_tst
current_time = datetime.now().strftime('%Y%m%d_%H%M%S')
sub_to_csv(sub, y_trn, trn_gini, etc07.__name__, current_time)
plot_importance(fscore, trn_gini, etc07.__name__, current_time)
# Global params
# ===============================================================
MAX_ROUND = 800
# ===============================================================
# Load and downcast
# ===============================================================
train, x_test, sub = load_all()
x_train = train.drop('target', axis=1)
y_train = train.target
x_train, x_test = downcast(x_train, x_test)
del train
gc.collect()
# ===============================================================
# Run cv and bags
# ===============================================================
n_splits = 5
folds = list(
KFold(n_splits=n_splits, shuffle=True,
random_state=99).split(x_train, y_train))
trn_gini, y_trn, y_tst, fscore = xgb03(x_train, y_train, x_test, folds,
MAX_ROUND, n_splits)
# ===============================================================
# Save submission and figure
# ===============================================================
if MAX_ROUND > 300:
sub.target = y_tst
current_time = datetime.now().strftime('%Y%m%d_%H%M%S')
sub_to_csv(sub, y_trn, trn_gini, xgb03.__name__, current_time)
plot_importance(fscore, trn_gini, xgb03.__name__, current_time)
# Global params
# ===============================================================
MAX_ROUND = 1200
# ===============================================================
# Load and downcast
# ===============================================================
train, x_test, sub = load_all()
x_train = train.drop('target', axis=1)
y_train = train.target
x_train, x_test = downcast(x_train, x_test)
del train
gc.collect()
# ===============================================================
# Run cv and bags
# ===============================================================
n_splits = 5
folds = list(
KFold(n_splits=n_splits, shuffle=True,
random_state=47).split(x_train, y_train))
trn_gini, y_trn, y_tst, fscore = cat05(x_train, y_train, x_test, folds,
MAX_ROUND, n_splits)
# ===============================================================
# Save submission and figure
# ===============================================================
if MAX_ROUND > 300:
sub.target = y_tst
current_time = datetime.now().strftime('%Y%m%d_%H%M%S')
sub_to_csv(sub, y_trn, trn_gini, cat05.__name__, current_time)
plot_importance(fscore, trn_gini, cat05.__name__, current_time)
# Load and downcast
# ===============================================================
train, x_test, sub = load_all()
x_train = train.drop('target', axis=1)
y_train = train.target
x_train, x_test = downcast(x_train, x_test)
del train
gc.collect()
# ===============================================================
# Run cv and bags
# ===============================================================
n_splits = 5
bags = 5
folds = list(
KFold(n_splits=n_splits, shuffle=True,
random_state=177).split(x_train, y_train)) # seed = 177
trn_gini, y_trn, y_tst, y_tst_mrank, fscore = xgb_bagging(
x_train, y_train, x_test, folds, MAX_ROUND, n_splits, bags)
# ===============================================================
# Save submission and figure
# ===============================================================
if MAX_ROUND > 100:
sub.target = y_tst
current_time = datetime.now().strftime('%Y%m%d_%H%M%S')
# sub_to_csv(sub, y_trn, trn_gini, "BadOfFiveMediumRank", current_time)
sub_to_csv(sub, y_tst_mrank, trn_gini, xgb_bagging.__name__,
current_time)
plot_importance(fscore, trn_gini, xgb_bagging.__name__, current_time)
# Global params
# ===============================================================
MAX_ROUND = 1200
# ===============================================================
# Load and downcast
# ===============================================================
train, x_test, sub = load_all()
x_train = train.drop('target', axis=1)
y_train = train.target
x_train, x_test = downcast(x_train, x_test)
del train
gc.collect()
# ===============================================================
# Run cv and bags
# ===============================================================
n_splits = 5
folds = list(
KFold(n_splits=n_splits, shuffle=True,
random_state=47).split(x_train, y_train))
trn_gini, y_trn, y_tst, fscore = log06(x_train, y_train, x_test, folds,
MAX_ROUND, n_splits)
# ===============================================================
# Save submission and figure
# ===============================================================
if MAX_ROUND > 300:
sub.target = y_tst
current_time = datetime.now().strftime('%Y%m%d_%H%M%S')
sub_to_csv(sub, y_trn, trn_gini, log06.__name__, current_time)
plot_importance(fscore, trn_gini, log06.__name__, current_time)
# Global params
# ===============================================================
MAX_ROUND = 1200
# ===============================================================
# Load and downcast
# ===============================================================
train, x_test, sub = load_all()
x_train = train.drop('target', axis=1)
y_train = train.target
x_train, x_test = downcast(x_train, x_test)
del train
gc.collect()
# ===============================================================
# Run cv and bags
# ===============================================================
n_splits = 5
folds = list(
KFold(n_splits=n_splits, shuffle=True,
random_state=47).split(x_train, y_train))
trn_gini, y_trn, y_tst, fscore = rgf04(x_train, y_train, x_test, folds,
MAX_ROUND, n_splits)
# ===============================================================
# Save submission and figure
# ===============================================================
if MAX_ROUND > 300:
sub.target = y_tst
current_time = datetime.now().strftime('%Y%m%d_%H%M%S')
sub_to_csv(sub, y_trn, trn_gini, rgf04.__name__, current_time)
plot_importance(fscore, trn_gini, rgf04.__name__, current_time)