def load_pkl(name):
"""Load xgboost model from pickle and perform conversion from version
0.90 if necessary.
:return:
XGBoost model
"""
import pickle
import xgboost
with open(name, 'rb') as f:
try:
model = pickle.load(f)
return model
except xgboost.core.XGBoostError as e:
if "Check failed: header == serialisation_header_" in str(e):
import xgboost_prev # pylint: disable=unused-import
import tempfile
class Unpickler(pickle.Unpickler):
def find_class(self, module, name):
if module.startswith("xgboost"):
return pickle.Unpickler.find_class(
self, module.replace("xgboost",
"xgboost_prev"), name)
return pickle.Unpickler.find_class(self, module, name)
f.seek(0)
model = Unpickler(f).load()
temp_file = tempfile.NamedTemporaryFile(
prefix='xgboost_migration', suffix='.model')
model.save_model(temp_file.name)
migrated_model = xgboost.XGBModel()
migrated_model.load_model(temp_file.name)
return migrated_model
raise
def train_xgb_model(subtrain_x, subtrain_y, validation_x, validation_y,
best_model_fname):
print(subtrain_x.shape)
params = {
'min_child_weight': 1,
'learning_rate': 0.03,
'colsample_bytree': 0.9,
'subsample': 0.9,
'gamma': 1,
'silent': 0,
'seed': 1234,
# 'booster': 'gblinear',
# 'booster': 'gbtree',
'max_depth': 9,
'objective': 'reg:linear',
'nthread': 10,
'n_estimators': 2000,
}
# xgsubtrain = xgb.DMatrix(subtrain_x, label=subtrain_y, )
# xgval = xgb.DMatrix(validation_x, label=validation_y)
# rgs = xgb.train(params, xgsubtrain, early_stopping_rounds=10, eval=(xgval, 'eval'))
rgs = xgb.XGBModel(**params)
rgs.fit(
subtrain_x,
subtrain_y,
eval_set=[(subtrain_x, subtrain_y), (validation_x, validation_y)],
eval_metric='mae',
early_stopping_rounds=30,
verbose=True,
)
return rgs, mean_absolute_error(validation_y, rgs.predict(validation_x))
def __init__(self):
self.preprocessor = joblib.load(os.getenv('PREPROCESSOR_PATH'))
self.clf = xgb.XGBModel(**{
'objective': 'binary:logistic',
'n_estimators': 10
})
self.clf.load_model(os.getenv('MODEL_PATH'))
def test_save_load_model():
tm._skip_if_no_sklearn()
from sklearn.datasets import load_digits
try:
from sklearn.model_selection import KFold
except:
from sklearn.cross_validation import KFold
digits = load_digits(2)
y = digits['target']
X = digits['data']
try:
kf = KFold(y.shape[0], n_folds=2, shuffle=True, random_state=rng)
except TypeError: # sklearn.model_selection.KFold uses n_split
kf = KFold(n_splits=2, shuffle=True,
random_state=rng).split(np.arange(y.shape[0]))
with TemporaryDirectory() as tempdir:
model_path = os.path.join(tempdir, 'digits.model')
for train_index, test_index in kf:
xgb_model = xgb.XGBClassifier().fit(X[train_index], y[train_index])
xgb_model.save_model(model_path)
xgb_model = xgb.XGBModel()
xgb_model.load_model(model_path)
preds = xgb_model.predict(X[test_index])
labels = y[test_index]
err = sum(1 for i in range(len(preds))
if int(preds[i] > 0.5) != labels[i]) / float(len(preds))
assert err < 0.1
def __init__(self, **kwargs):
self.n_estimators = kwargs['xgb_n_estimators']
self.objective = kwargs['xgb_objective']
self.eval_metric = kwargs['xgb_eval_metric']
self.verbose = kwargs['xgb_verbose']
self.model = xgb.XGBModel(n_estimators=self.n_estimators,
objective=self.objective)
def save_load_model(model_path):
from sklearn.datasets import load_digits
from sklearn.model_selection import KFold
digits = load_digits(2)
y = digits['target']
X = digits['data']
kf = KFold(n_splits=2, shuffle=True, random_state=rng)
for train_index, test_index in kf.split(X, y):
xgb_model = xgb.XGBClassifier().fit(X[train_index], y[train_index])
xgb_model.save_model(model_path)
xgb_model = xgb.XGBClassifier()
xgb_model.load_model(model_path)
assert isinstance(xgb_model.classes_, np.ndarray)
assert isinstance(xgb_model._Booster, xgb.Booster)
assert isinstance(xgb_model._le, XGBoostLabelEncoder)
assert isinstance(xgb_model._le.classes_, np.ndarray)
preds = xgb_model.predict(X[test_index])
labels = y[test_index]
err = sum(1 for i in range(len(preds))
if int(preds[i] > 0.5) != labels[i]) / float(len(preds))
assert err < 0.1
assert xgb_model.get_booster().attr('scikit_learn') is None
# test native booster
preds = xgb_model.predict(X[test_index], output_margin=True)
booster = xgb.Booster(model_file=model_path)
predt_1 = booster.predict(xgb.DMatrix(X[test_index]),
output_margin=True)
assert np.allclose(preds, predt_1)
with pytest.raises(TypeError):
xgb_model = xgb.XGBModel()
xgb_model.load_model(model_path)
def search_cv(x_train, y_train, x_test):
xgb_model = xgb.XGBModel()
params = {
'booster': ['gblinear'],
'silent': [1],
'learning_rate': [x for x in np.round(np.linspace(0.01, 1, 20), 2)],
'reg_lambda': [lambd for lambd in np.logspace(0, 3, 50)],
'objective': ['reg:linear']
}
print('begin')
clf = GridSearchCV(xgb_model,
params,
scoring='neg_mean_squared_error',
refit=True)
clf.fit(x_train, y_train)
preds = clf.predict(x_test)
sub_df = pd.read_csv('raw_data/answer_sample_b_20180117.csv', header=None)
sub_df['Value'] = preds
sub_df.to_csv('result/xgboost4.csv', header=None, index=False)
best_parameters, score, _ = max(clf.grid_scores_, key=lambda x: x[1])
print('Raw RMSE:', score)
for param_name in sorted(best_parameters.keys()):
print("%s: %r" % (param_name, best_parameters[param_name]))
def fit(self):
#print(self.args, self.X.shape, self.y.shape)
appFeatures = ["%s_%d_%d" % (x, i, t) for t in [1, 0] for x in libdata.apprates for i in [0,1]]
phyFeatures = ["%s_0" % (x) for x in libdata.targets]
feature_names = np.append(appFeatures, phyFeatures)
X = self.X
y = self.y
#print(X)
if self.m is not None and self.m < len(self.X):
sels = random.sample(X.shape[0], self.m, replace = False)
X = X[sels]
y = y[sels]
self.nys = y.shape[1]
print(len(X))
for i in range(self.nys):
self.models.append(xgb.XGBModel(**self.args))
self.models[i].fit(X, y[:,i])
if False:
#plot_importance(self.models[i])
feature_importance = self.models[i].feature_importances_
feature_importance = 100.0 * (feature_importance / feature_importance.max())
sorted_idx = np.argsort(feature_importance)
sorted_idx = sorted_idx[-10:]
pos = np.arange(sorted_idx.shape[0]) + .5
plt.figure()
plt.barh(pos, feature_importance[sorted_idx], align='center')
print(sorted_idx)
plt.yticks(pos, feature_names[sorted_idx])
plt.xlabel('Relative Importance')
plt.title('Feature Importance for Fan Power Prediction')
plt.savefig("power_10r.eps", bbox_inches='tight')
print(feature_importance)
def test_save_load_model():
with TemporaryDirectory() as tempdir:
model_path = os.path.join(tempdir, 'digits.model')
save_load_model(model_path)
with TemporaryDirectory() as tempdir:
model_path = os.path.join(tempdir, 'digits.model.json')
save_load_model(model_path)
from sklearn.datasets import load_digits
with TemporaryDirectory() as tempdir:
model_path = os.path.join(tempdir, 'digits.model.json')
digits = load_digits(2)
y = digits['target']
X = digits['data']
booster = xgb.train(
{
'tree_method': 'hist',
'objective': 'binary:logistic'
},
dtrain=xgb.DMatrix(X, y),
num_boost_round=4)
predt_0 = booster.predict(xgb.DMatrix(X))
booster.save_model(model_path)
cls = xgb.XGBClassifier()
cls.load_model(model_path)
predt_1 = cls.predict(X)
assert np.allclose(predt_0, predt_1)
cls = xgb.XGBModel()
cls.load_model(model_path)
predt_1 = cls.predict(X)
assert np.allclose(predt_0, predt_1)
def XGBoost(self):
self.Encoding()
X_train,X_test,y_train,y_test = train_test_split(Train.X,Train.y,test_size=0.25,random_state=4)
clf = xgb.XGBModel(max_depth=8,n_estimators=100,objective="reg:linear", random_state=17,n_jobs=-1)
clf.fit(X_train, y_train, eval_metric='rmse', verbose = True, eval_set = [(X_train,y_train),(X_test, y_test)])
clf.save_model('./model/XGBoost.model')
pickle.dump(clf, open("XGBosst.pickle.dat", "wb"))
def predict_cy_young(data_path, model_name, labels_to_drop=None):
bst = xgb.XGBModel()
bst.load_model(model_name)
data = pd.read_csv(data_path)
to_predict = data.drop(labels=labels_to_drop, axis=1)
ypreds = bst.predict(to_predict)
return ypreds
def __init__(self,
task: Task,
scorer: Scorer,
opt_logger: OptimizationLogger = VoidLogger(None)):
if task.task == "classification":
space = XGBoostOptimizer.Params.classification_space
else:
space = XGBoostOptimizer.Params.general_space
super().__init__(xgb.XGBModel(), task, space, scorer, opt_logger)
def train_xgboost(data, avg={}):
test_X, test_Y = load_data_no_cut(data, avg)
bst = xgb.XGBModel(max_depth=6,
learning_rate=0.1,
silent=True,
objective='reg:linear',
subsample=0.7,
reg_alpha=0.5,
reg_lambda=0.3,
n_estimators=80)
# bst.set_params(**param)
bst.fit(test_X, test_Y)
return bst
def xgboosting(X_train, y_train, n_estimators, params):
print("> Model type : XGBoost")
model = xgb.XGBModel(objective='reg:squarederror',
max_depth=11,
subsample=0.5,
colsample_bytree=0.5,
learning_rate=0.1,
n_estimators=n_estimators,
verbosity=0,
seed=42)
power_lines = y_train.columns
trained_models = {}
for pl in power_lines:
trained_model = model.fit(X_train, y_train[pl])
trained_models[pl] = trained_model
save_model(trained_models, "xgboost", False, params)
def train_xgboost(train_x, train_y, test_x, test_y, data_c):
param = {"booster":"gbtree", "max_depth": 2, "eta": 0.3, "objective": "binary:logistic", "nthread":2}
num_round = 100
train_mat = xgb.DMatrix(train_x, train_y)
test_mat = xgb.DMatrix(test_x, label=test_y)
all_mat = xgb.DMatrix(data_c.drop(columns=["Prod1"]), label=data_c[["Prod1"]])
evaluation = [(test_mat, "eval"), (train_mat, "train")]
bst = xgb.train(param, train_mat, num_round, evaluation)
clf3 = xgb.XGBModel(**param)
clf3.fit(train_x, train_y, eval_set=[(train_x, train_y), (test_x, test_y)], eval_metric='logloss')
print(roc_auc_score(test_y, bst.predict(test_mat)))
print(roc_auc_score(test_y, clf3.predict(test_x)))
def test_save_load_model(self):
self._init_ray()
with TemporaryDirectory() as tempdir:
model_path = os.path.join(tempdir, "digits.model")
self.save_load_model(model_path)
with TemporaryDirectory() as tempdir:
model_path = os.path.join(tempdir, "digits.model.json")
self.save_load_model(model_path)
from sklearn.datasets import load_digits
with TemporaryDirectory() as tempdir:
model_path = os.path.join(tempdir, "digits.model.json")
digits = load_digits(n_class=2)
y = digits["target"]
X = digits["data"]
booster = xgb.train(
{
"tree_method": "hist",
"objective": "binary:logistic"
},
dtrain=xgb.DMatrix(X, y),
num_boost_round=4,
)
predt_0 = booster.predict(xgb.DMatrix(X))
booster.save_model(model_path)
cls = RayXGBClassifier()
cls.load_model(model_path)
proba = cls.predict_proba(X)
assert proba.shape[0] == X.shape[0]
assert proba.shape[1] == 2 # binary
predt_1 = cls.predict_proba(X)[:, 1]
assert np.allclose(predt_0, predt_1)
cls = xgb.XGBModel()
cls.load_model(model_path)
predt_1 = cls.predict(X)
assert np.allclose(predt_0, predt_1)
def test_save_load_model():
from sklearn.datasets import load_digits
from sklearn.model_selection import KFold
digits = load_digits(2)
y = digits['target']
X = digits['data']
kf = KFold(n_splits=2, shuffle=True, random_state=rng)
with TemporaryDirectory() as tempdir:
model_path = os.path.join(tempdir, 'digits.model')
for train_index, test_index in kf.split(X, y):
xgb_model = xgb.XGBClassifier().fit(X[train_index], y[train_index])
xgb_model.save_model(model_path)
xgb_model = xgb.XGBModel()
xgb_model.load_model(model_path)
preds = xgb_model.predict(X[test_index])
labels = y[test_index]
err = sum(1 for i in range(len(preds))
if int(preds[i] > 0.5) != labels[i]) / float(len(preds))
assert err < 0.1
def best_xgb(X_train, y_train, n_estimators, params):
models = {}
importance_tab = load("importance_per_w")
power_lines = y_train.columns[y_train.columns.str.match("NPWD")]
model = xgb.XGBModel(objective='reg:squarederror',
max_depth=11,
subsample=0.5,
colsample_bytree=0.5,
learning_rate=0.1,
n_estimators=n_estimators,
verbosity=0,
seed=42)
trained_models = {}
for pl in power_lines:
print(">> Fitting", pl)
features = importance_tab[pl]
model.fit(X_train[features[:40]], y_train[pl])
trained_models[pl] = model
save_model(trained_models, "xgboost", False, params)
def search(self, x_train, y_train, x_test):
xgb_model = xgb.XGBModel()
params = {
'booster': ['gblinear'],
'silent': [1],
'learning_rate':
[x for x in np.round(np.linspace(0.01, 1, 20), 2)],
'reg_lambda': [lambd for lambd in np.logspace(0, 3, 50)],
'objective': ['reg:linear']
}
print('begin')
clf = GridSearchCV(xgb_model,
params,
scoring='neg_mean_squared_error',
refit=True)
clf.fit(x_train, y_train)
preds = clf.predict(x_test)
return preds
def get_estimator(estimator):
if estimator == 'booster':
e = xgb.XGBModel(objective='reg:squarederror',
max_depth=11,
subsample=0.5,
colsample_bytree=0.5,
learning_rate=0.1,
n_estimators=500,
silent=1,
seed=42)
elif estimator == 'xtrees':
e = ExtraTreesRegressor(n_estimators=500,
random_state=0,
min_samples_leaf=20,
n_jobs=-1)
elif estimator == 'rf':
e = RandomForestRegressor(n_estimators=500,
random_state=1,
min_samples_leaf=10,
n_jobs=-1)
return e
def train_with_xgboost(self, x_train, y_train, x_test, y_test):
xgb_model = xgb.XGBModel()
params = {
'booster': ['gblinear'],
'silent': [1],
'learning_rate':
[x for x in np.round(np.linspace(0.01, 1, 20), 2)],
'reg_lambda': [lambd for lambd in np.logspace(0, 3, 50)],
'objective': ['reg:linear']
}
print('begin')
clf = GridSearchCV(xgb_model,
params,
scoring='neg_mean_squared_error',
refit=True)
clf.fit(x_train, y_train)
preds = clf.predict(x_test)
print('test mse:', self.cal_MSE(preds, y_test))
best_parameters, score, _ = max(clf.grid_scores_, key=lambda x: x[1])
print('Raw RMSE:', score)
for param_name in sorted(best_parameters.keys()):
print("%s: %r" % (param_name, best_parameters[param_name]))
def test_multi_adboost_cart(data):
test_X, test_Y = load_data(data)
adaboost = MultiAdaBoostRegressor([
DecisionTreeRegressor(max_depth=4),
GradientBoostingRegressor(n_estimators=1,
learning_rate=0.1,
max_depth=4,
random_state=0,
loss='ls'),
xgb.XGBModel(max_depth=4,
learning_rate=0.6,
silent=True,
objective='reg:linear',
subsample=0.7,
reg_alpha=0.5,
reg_lambda=0.3,
n_estimators=1)
],
loss="square",
learning_rate=0.01,
n_estimators=4)
adaboost.fit(test_X, test_Y)
return adaboost
# -*- coding: utf-8 -*-
# @Time : 2019/3/15 15:36
# @Author : lilong
# @File : xgboost_intro.py
# @Description: xgboost简介
# 多分类问题指定 objective为'multi:softmax'
import numpy as np
import xgboost as xgb
xgb.XGBModel()
if __name__ == '__main__':
data_train = xgb.DMatrix(r'E:\pyProject\python_trick\xgboost_practice\data\14.agaricus_train.txt')
data_test = xgb.DMatrix(r'E:\pyProject\python_trick\xgboost_practice\data\14.agaricus_test.txt')
print(data_train)
print(data_test)
# 设置参数
param = {'max_depth': 2, 'eta': 1, 'silent': 1, 'objective': 'binary:logistic'} # logitraw
# xgb.XGBClassifier()
watchlist = [(data_test, 'eval'), (data_train, 'train')]
bst = xgb.train(param, data_train, num_boost_round=3, evals=watchlist)
# bst = xgb.train(param, data_train, num_boost_round=n_round, evals=watchlist, obj=log_reg, feval=error_rate)
y_pred = bst.predict(data_test)
y = data_test.get_label()
error = sum(y != (y_pred > 0.5))
err_rate = float(error) / len(y_pred)
print('错误数据%d' % error)
print('错误率%.5f%%' % (err_rate * 100))
def main(args, path: str, prefix: str, model_path: str, date_str: str):
input_col = "{0}/{1}_columns.npy".format(path, prefix)
input_x_train = "{0}/{1}_x_train.npy".format(path, prefix)
input_y_train = "{0}/{1}_y_train.npy".format(path, prefix)
input_x_dev = "{0}/{1}_x_dev.npy".format(path, prefix)
input_y_dev = "{0}/{1}_y_dev.npy".format(path, prefix)
model_file = "{0}/{1}_({2}).model".format(model_path, date_str, prefix)
config_file = "{0}/{1}_({2}).config".format(model_path, date_str, prefix)
importances_file = "{0}/{1}_({2}).xlsx".format(model_path, date_str,
prefix)
if args.configuration is None:
config = {
"max_depth": 3,
"n_estimators": 400,
"min_child_weight": 1,
"tree_method": "gpu_hist",
"learning_rate": 0.07,
}
else:
with open(file=args.configuration, mode='r') as json_file:
config = json.load(json_file)
print("Start training...")
columns = np.load(input_col)
X_train = np.load(input_x_train)
Y_train = np.load(input_y_train)
if args.split_train:
X_train, X_dev, Y_train, Y_dev = train_test_split(
X_train,
Y_train,
test_size=args.test_size,
stratify=Y_train,
)
else:
X_dev = np.load(input_x_dev)
Y_dev = np.load(input_y_dev)
# Imbalance ratio.
# Allows to compensate the imbalance between the classes.
imbalance_ratio = len(Y_train) / np.sum(Y_train)
print("Imbalance:", imbalance_ratio)
config.update({"scale_pos_weight": imbalance_ratio})
# Train the classifier
# "rmse" for root mean squared error.
# "mae" for mean absolute error.
# "logloss" for binary logarithmic loss
# and "mlogloss" for multi-class log loss (cross entropy).
# "error" for classification error.
# "auc" for area under ROC curve.
with open(file=config_file, mode='w+') as cfg_file:
json.dump(config, cfg_file)
clf = xgb.XGBModel(**config)
clf.fit(
X_train,
Y_train,
eval_set=[(X_dev, Y_dev)],
eval_metric=["auc"],
early_stopping_rounds=400,
verbose=True,
)
bst = clf.get_booster()
bst.save_model(model_file)
imp = clf.feature_importances_
for feature_name, importance in zip(imp, columns):
print(
"Feature name: {0}, importance: {1}".format(
importance,
feature_name,
), )
# Output the features importance in an Excel file
df_importances = pd.DataFrame(
np.hstack((
np.array(
sorted(
zip(columns, imp, imp / np.max(imp)),
key=lambda line: line[1],
reverse=True,
), ),
np.array([np.cumsum(sorted(imp, reverse=True))]).transpose(),
), ),
columns=[
'Feature name',
'Importance',
'Normalised importance',
'Cumulated importance',
],
)
df_importances['Importance'] = pd.to_numeric(df_importances['Importance'])
df_importances['Normalised importance'] = pd.to_numeric(
df_importances['Normalised importance'], )
df_importances['Cumulated importance'] = pd.to_numeric(
df_importances['Cumulated importance'], )
with pd.ExcelWriter( # pylint: disable=abstract-class-instantiated
importances_file,
date_format='YYYY-MM-DD',
datetime_format='YYYY-MM-DD HH:MM:SS',
engine='xlsxwriter',
) as writer:
df_importances.to_excel(
writer,
sheet_name='Features importance',
index=False,
)
# https://stackoverflow.com/a/40535454
worksheet = writer.sheets['Features importance']
for idx, col in enumerate(df_importances): # loop through all columns
series = df_importances[col]
max_len = max(
(
series.astype(str).map(len).max(), # len of largest item
len(str(series.name)), # len of column name/header
), ) + 2 # adding a little extra space
worksheet.set_column(idx, idx, max_len) # set column width
# @Time : 2019/3/18 18:44
# @Author : lilong
# @File : xgboost_Model.py
# @Description:
import xgboost as xgb
from sklearn.datasets import load_iris
if __name__ == '__main__':
data_train = xgb.DMatrix(r'E:\pyProject\python_trick\xgboost_practice\data\14.agaricus_train.txt')
data_test = xgb.DMatrix(r'E:\pyProject\python_trick\xgboost_practice\data\14.agaricus_test.txt')
iris_data = load_iris()
X = iris_data.data
y = iris_data.target
# 设置参数
param = {'max_depth': 2, 'eta': 1, 'silent': 1, 'objective': 'binary:logistic'} # logitraw
bst = xgb.XGBModel(objective="reg:linear",
booster='gbtree',
max_depth=3,
learning_rate=1,
n_estimators=4
)
bst.fit(X, y)
y_pred = bst.predict(X)
y = data_test.get_label()
error = sum(y != (y_pred > 0.5))
err_rate = float(error) / len(y_pred)
print('错误数据%d' % error)
print('错误率%.5f%%' % (err_rate * 100))
import pandas as pd
import numpy as np
from math import exp
from sklearn.preprocessing import normalize
import xgboost as xgb
norms = pd.read_excel('data.xlsx', sheet_name='normatives', index_col='Номер')
clf = xgb.XGBModel()
clf.load_model('xgb_class.json')
def to_float(x):
try:
x = x.replace(',', '.')
except:
pass
return float(x)
for item in ['Зона опасности', 'Зона риска', 'Зона стабильности']:
norms[item] = norms[item].apply(to_float)
def belong_f(x, param=1):
"""
Функция, возвращающая кортеж значений функций принадлежности к каждому из состояний
"""
# значения, разделяющие интервалы
splitters = [
norms.at[param, 'Зона риска'], norms.at[param, 'Зона опасности'],
norms.at[param, 'Зона стабильности']
dtest = xgb.DMatrix(X_test, label=y_test)
# param = [('max_depth', 2), ('objective', 'binary:logistic'), ('eval_metric', 'logloss'), ('eval_metric', 'error'), ('n_estimators',2)]
param = {'max_depth': 2, 'objective': 'binary:logistic', 'eval_metric': ['logloss', 'error'], 'n_estimators': 2}
num_round = 2
watchlist = [(dtest,'eval'), (dtrain,'train')]
evals_result = {}
bst = xgb.train(param, dtrain, num_round, watchlist, evals_result=evals_result)
print('Access logloss metric directly from evals_result:')
print(evals_result['eval']['logloss'])
print('')
print('Access complete dictionary:')
print(evals_result)
"""
param_dist = {'objective': 'binary:logistic', 'n_estimators': 2}
clf = xgb.XGBModel(**param_dist)
clf.fit(X_train, y_train,
eval_set=[(X_train, y_train), (X_test, y_test)],
eval_metric=['logloss', 'error'],
verbose=True)
evals_result = clf.evals_result()
print(evals_result)
import os
import sys
import joblib
import xgboost as xgb
import pandas as pd
from flask import Flask
from flask import request
from flask.logging import default_handler
app = Flask(__name__)
preprocessor = joblib.load(os.getenv('PREPROCESSOR_PATH'))
clf = xgb.XGBModel(**{'objective': 'binary:logistic', 'n_estimators': 10})
clf.load_model(os.getenv('MODEL_PATH'))
@app.route('/predict', methods=['POST'])
def predict():
requestJSON = request.get_json(force=True, cache=False)
features = pd.DataFrame({
'PassengerId': pd.Series([], dtype='int64'),
'Survived': pd.Series([], dtype='int64'),
'Pclass': pd.Series([], dtype='int64'),
'Name': pd.Series([], dtype='str'),
'Sex': pd.Series([], dtype='str'),
'Age': pd.Series([], dtype='float64'),
'SibSp': pd.Series([], dtype='int64'),
'Parch': pd.Series([], dtype='int64'),
'Ticket': pd.Series([], dtype='str'),
def run(k, j, filename, seednum=20, threshold = 0.5, resultdir=None, graphdir = f'{treedir}/'):
# classes = ["P1a1" , "P1a2" , "P2b" , "P2c" ]
classes = ["P1a1" , "P1a2", "P2b", "P2c", "H1" ]
# H1 H2 O (1) P1a1 (4) P1a2 (6) P2b P2c S1a (0) S1c S2 S3
joind = gp.read_file(filename, layer = layers[j])
print(f'\n------\n------{layers[j]}----\n-----\n')
joind['area']= joind['geometry'].area #calculate the area of each object
df1 = pd.DataFrame(joind.drop(columns='geometry'))
df1 = df1.replace([np.inf, -np.inf], np.nan).dropna()
Pcl = df1.loc[df1['geocode_2'].isin(classes)] # filter only classes of interest
print(Pcl['geocode_2'].value_counts())
# regroup, geocode_2 from here on becomes binary!
Pcl['geocode_2'] = np.where(Pcl['geocode_2'].str.contains(classes[k]),classes[k],'Others')
print(Pcl['geocode_2'].value_counts())
minc = min(Pcl['geocode_2'].value_counts() ) # skip if less than 20 objects
if minc< 20:
print("minimum class less than 20")
return (-1, -1) # -1 -1 if not calculated
else:
print(f'total {len(df1)}, P_H1_classes: {len(Pcl)}, minimun class: {minc}')
# bootstrap and get averaged accuracy
avepre = np.zeros(1) # store all the xgb+tree precisions in each CV
averec = np.zeros(1)
for seeds in range(seednum):
np.random.seed(seeds)
#1. categorise the variable "area", the variable "area" is kept in the data frame, strictly it can be removed.
#2. use groupby to sample the same amount for each area category
# use 70% of area for training, get the index
print (Pcl['area'].quantile([0, .25, .5, .75, 1]))
Pcl['area_c'] = pd.cut(Pcl['area'],
bins= Pcl['area'].quantile([0, .25, .5, .75, 1]).tolist()
labels=[ "q25", "q5", "q75", "Max"])
print(Pcl["area_c"].value_counts())
train_ind = Pcl.groupby('area_c').sample(n = int(min(Pcl["area_c"].value_counts())*0.7)).index
test_ind = Pcl[~Pcl.index.isin(train_ind)].index
Pcl.loc [train_ind,"geocode_2" ].value_counts()
X_train0 = Pcl.loc [train_ind ].drop(columns=["geocode_2","layer","OBJECTID","path", "area_c"])
X_test0 = Pcl.loc [test_ind ].drop(columns=["geocode_2","layer","OBJECTID","path", "area_c"])
Y_train0 = Pcl.filter(regex='geocode_2').loc[train_ind]
Y_test0 = Pcl.filter(regex='geocode_2').loc[test_ind]
print("after sampling by area: for 2 classes,", X_train0.shape[0], X_test.shape[0])
print(Pcl.loc [train_ind ]["geocode_2"].value_counts())
# if my pandas is lower and i can't use the above function,
# grouped = Pcl.drop(columns=["geocode_2","layer","OBJECTID","path",'area']).groupby('area_c')
#def fun1(x):
# y = x.drop(columns=["area_c"])
# return( y.sample(n = int(minc/5*0.7)).index )
#train_ind = grouped.apply(fun1)
#test_ind = Pcl[~Pcl.index.isin(train_ind)].index
#neew to ungroup train_ind
# test data
#grouped2 = Pcl[['geocode_2',"area_c"]].groupby('area_c')
#y = grouped2.apply(fun1)
#####
# after getting x, y train, we will use undersample to sample from each classes, p1a1 and others
rus = RandomUnderSampler(random_state = 1)
X_train, Y_train = rus.fit_resample(X_train0, Y_train0)
print("number of samples used for training:", X_train.shape[0]/2)
#y2 = y2.reshape(-1, 1)
#y2_rus, y_rus = rus.fit_resample(y2, y)
#y2_rus= y2_rus.flatten()
#len(train)+len(test)
# relable
label_all = [classes[k], "Others"]
#classtype = [(j, "float32") for j in classes]
#Pcl.geocode_2.unique()
i = 0
idx2class = {}
class2idx = {}
for tp in label_all:
idx2class[i] = tp
class2idx[tp] = i
i+= 1
Y_trainnum = cl2idx(Y_train.values, class2idx).astype(int)
Y_testnum = cl2idx(Y_test.values, class2idx).astype(int)
np.unique(Y_trainnum)
params = {'max_depth': 6, 'eta': 0.002,
'objective':'binary:logistic', 'num_class': 1}
clf = xgb.XGBModel(**params)
clf.fit(X_train.values, Y_trainnum,
eval_set=[(X_train.values, Y_trainnum), (X_test.values, Y_testnum)],
eval_metric='logloss',
verbose=True)
#for testing
#clf = DecisionTreeClassifier(min_samples_split= 30, max_depth= 4, min_samples_leaf=20, random_state=1)
yhat = clf.predict(X_test)
# threshold 0.5, probability higher than 0.5 -> positive.
yhat_labels = yhat>threshold
yhat_labels = yhat_labels.astype(int)
#TP
TP = ((Y_testnum == 1) & (yhat_labels == 1)).astype(float) * X_test["area"]
#FP
FP = ((Y_testnum == 0) & (yhat_labels == 1)).astype(float) * X_test["area"]
#TN
TN = ((Y_testnum == 0) & (yhat_labels == 0)).astype(float) * X_test["area"]
#FN
FN =((Y_testnum == 1) & (yhat_labels == 0)).astype(float) * X_test["area"]
precision = np.sum(TP)/np.sum(TP+FP)
recall = np.sum(TP)/np.sum(TP+TN)
averec = np.append(averec, recall) #store all of them
avepre = np.append(avepre, precision)
recall = averec.sum()/seednum #get the mean but exclude the first one (0)
precision = avepre.sum()/seednum
print(averec, recall)
if resultdir is not None:
Y_testnum = Y_testnum.astype(int)
plt.rcParams.update({'font.size': 8})
ax = xgb.plot_importance(model, grid=False, importance_type='gain', title='Feature importance')
ax.set_title(f'xgboost importance {layers[j]} {classes[k]}')
fname = f"{resultdir}/P_{layers[j]}_{classes[k]}_imp"
plt.savefig(fname, dpi=1200)
return (recall, precision)
# Boosted tree
print("XGB Tree")
import xgboost as xgb
param = {
'n_estimators': 10000,
'learning_rate': 0.1,
'objective': 'reg:squarederror',
'verbosity': 0
}
fit_param = {
'eval_set': [(X_train, y_train), (X_test, y_test)],
'early_stopping_rounds': 200,
'verbose': False
}
BT = Regressor(xgb.XGBModel(**param))
BT.run(X_train, y_train, X_test, y_test, **fit_param)
X_train_new, X_test_new = BT.select(X_train, X_test)
fit_param = {
'eval_set': [(X_train_new, y_train), (X_test_new, y_test)],
'early_stopping_rounds': 100,
'verbose': False
}
# BT.run(X_train_new, y_train, X_test_new, y_test, **fit_param)
xgb.plot_importance(BT.reg)
plt.show()
# RF
print("Random Forest")
RF = Regressor(ensemble.RandomForestRegressor(random_state=42))
RF.run(X_train, y_train, X_test, y_test)