def test_sklearn_deploy(self, mock_post, mock_put):
model_name = 'test-model'
mock_post_response = Mock()
mock_post_response.json.return_value = {'url': 'http://test-url', 'tier': 0}
mock_post_response.status_code = 200
mock_post.return_value = mock_post_response
mock_put_response = Mock()
mock_put_response.json.return_value = {}
mock_put_response.status_code = 200
mock_put.return_value = mock_put_response
if path.exists(model_name):
clf = skljson.from_json(model_name)
else:
X, y = make_classification(n_samples=50, n_features=3, n_classes=3, n_informative=3, n_redundant=0,
random_state=0, shuffle=False)
clf = RandomForestClassifier()
clf.fit(X, y)
skljson.to_json(clf, model_name)
sklearn = SKLearn('test')
sklearn.deploy(clf, model_name)
mock_post.assert_called_once()
mock_put.assert_called_once()
def deserialize_model(path):
# Load (or deserialize) model from JSON
model = skljson.from_json(path)
# Convert coeficients to numpy arrays to enable JSON deserialization
# This is a hack to compensate for a bug in sklearn_json
for i, x in enumerate(model.coefs_):
model.coefs_[i] = np.array(x)
return model
def __init__(self, modelFile, game, config="configs/config.ini"):
ext = modelFile.split(sep='.')[-1]
self.model = None
self.game = game
if (config == None):
self.config = "configs/config.ini"
else:
self.config = config
if (ext == "json"):
self.model = (skljson.from_json(modelFile))
elif (ext == "pickle"):
with open(modelFile, 'rb') as f:
self.model = pickle.load(f)
else:
print("Unknown extention")
def check_sparse_model_json(self, model, model_name, abs=False):
# Given
if abs:
model.fit(np.absolute(self.X_sparse), self.y_sparse)
else:
model.fit(self.X_sparse, self.y_sparse)
# When
serialized_model = skljson.to_json(model, model_name)
deserialized_model = skljson.from_json(model_name)
# Then
expected_predictions = model.predict(self.X)
actual_predictions = deserialized_model.predict(self.X)
testing.assert_array_equal(expected_predictions, actual_predictions)
def predict(
path="",
model=None,
prefix="",
out_dir="",
):
"""
----
path: str
Path to serialized/pickeld training set
model: sklearn model
Path to serialized RandomForest classifier (trained)
prefix:
prefix for .csv file with prediction results and serialized model
out_dir:
path to create output folder.
"""
if model is None:
model = Path(__file__).parent / "data/model/UP000005640_9606_llps.json"
try:
logger.info("Loading model: {m}", m=model)
clf = skljson.from_json(model)
except Exception:
logger.error("classifier {mod} not found. Does the file exist?",
mod=model)
mat = export_matrix(prefix=prefix, fasta_path=path, out_path=out_dir)
# Preprocessing
data_ps = preprocess_and_scaledata(mat.df)
X = data_ps.select_dtypes([np.number])
logger.info("Predicting PSAP_score")
psap_prediction = pd.DataFrame(index=data_ps["protein_name"])
psap_prediction["PSAP_score"] = clf.predict_proba(X)[:, 1]
psap_prediction["rank"] = 0
rank = psap_prediction["PSAP_score"].rank(ascending=False)
psap_prediction["rank"] = rank
# # Make directory for output
out_dir = Path(out_dir)
out_dir.mkdir(parents=True, exist_ok=True)
out_file = out_dir / f"prediction_{prefix}.csv"
logger.info("Writing results to: {csv}", csv=out_file)
psap_prediction.to_csv(out_file)
def test_sklearn_model_exceeded(self, mock_post):
model_name = 'test-model-1mb'
mock_post_response = Mock()
mock_post_response.json.return_value = {'url': 'http://test-url', 'tier': 0}
mock_post_response.status_code = 200
mock_post.return_value = mock_post_response
if path.exists(model_name):
clf = skljson.from_json(model_name)
else:
X, y = make_classification(n_samples=15000, n_features=10, n_classes=3, n_informative=3, n_redundant=0,
random_state=0, shuffle=False)
clf = RandomForestClassifier()
clf.fit(X, y)
skljson.to_json(clf, model_name)
sklearn = SKLearn('test')
with self.assertRaises(mlrequest.ModelSizeExceeded) as exception:
sklearn.deploy(clf, model_name)
mock_post.assert_called_once()
def load_serialized(filename_with_path):
""" load a serialized model """
if not os.path.isfile(filename_with_path):
print(
"{} is not a valid file, please check".format(filename_with_path))
return
feature_selection_filename = filename_with_path.replace(
"_reg.json", "_fs.json")
feature_selection = None
if os.path.isfile(feature_selection_filename):
feature_selection = json_load(feature_selection_filename)
regressor = skljson.from_json(filename_with_path)
# override n_jobs to prevent warning, model should be fast enough
# n_jobs helps during training
regressor.n_jobs = 1
class Model:
""" wrapper to the serialized scikit learn model,
that uses feature selection in the first step
"""
def __init__(self, regressor, fs=None):
self._regressor = regressor
self._fs = fs
def feature_select(self, X):
fs = np.array(self._fs)
X = np.array(X)
_X = []
# perform selection for each input row
for x in X:
_X.append(x[fs])
return _X
def predict(self, X):
if self._fs:
X = self.feature_select(X)
return self._regressor.predict(X)
return Model(regressor, feature_selection)
def conOutFinal(cfile, order):
a = open(cfile.replace('.tocnf', '.possum'), 'w')
try:
specMODEL = skljson.from_json(
pkg_resources.resource_filename(
'POSMM', 'models/species_' + str(order) + '.json'))
genMODEL = skljson.from_json(
pkg_resources.resource_filename(
'POSMM', 'models/genus_' + str(order) + '.json'))
famMODEL = skljson.from_json(
pkg_resources.resource_filename(
'POSMM', 'models/family_' + str(order) + '.json'))
claMODEL = skljson.from_json(
pkg_resources.resource_filename(
'POSMM', 'models/class_' + str(order) + '.json'))
ordMODEL = skljson.from_json(
pkg_resources.resource_filename(
'POSMM', 'models/order_' + str(order) + '.json'))
phyMODEL = skljson.from_json(
pkg_resources.resource_filename(
'POSMM', 'models/phylum_' + str(order) + '.json'))
except:
print('ERROR: Models Missing. Aborting')
sys.exit()
modlist = [phyMODEL, claMODEL, ordMODEL, famMODEL, genMODEL, specMODEL]
matchpos = []
for m in modlist:
mp = (list(m.classes_)).index('Match')
matchpos.append(mp)
with open(cfile) as infile:
for lines in infile:
lines = lines.rstrip()
values = lines.split('\t')
taxid = values[2]
lineage = glin[taxid][1:]
rawscore = float(values[1])
readlen = int(values[0])
wrstr = []
for x in range(len(modlist)):
cfs = modlist[x].predict_proba([[rawscore,
readlen]])[0][matchpos[x]]
wrstr.append(lineage[x] + ':::' + str(cfs))
a.write('\t'.join(wrstr) + '\n')
a.close()
def open_model(model_path):
"""Open and return a model from json file
:param model_path: path to the model
"""
if model_path.endswith('gz'):
with gzip.open(model_path, 'r') as f:
model = f.read()
model = json.loads(model.decode('utf-8'))
model = from_dict(model)
return model
else:
with open(model_path, 'r') as f:
a = f.readline()
if a.startswith('{"learner"'):
model = xgb.Booster()
model.load_model(model_path)
return model
else:
model = from_json(model_path)
return model
def predict(path, model, prefix="", out_dir=""):
"""
----
path: str
Path to serialized/pickeld training set
model: sklearn model
Path to serialized RandomForest classifier (trained)
prefix:
prefix for .csv file with prediction results and serialized model
out_dir:
path to create output folder.
"""
print("Loading model")
print(model)
try:
clf = skljson.from_json(model)
except:
print("An error occured while importing the model from json")
print("annotating fasta")
data = export_matrix(name=prefix, fasta_path=path, out_path=out_dir)
# Preprocessing
data_ps = preprocess_and_scaledata(data, "llps")
data_numeric = data_ps.select_dtypes([np.number])
X = data_numeric.drop("llps", axis=1)
y = data_numeric["llps"]
psap_prediction = pd.DataFrame(index=data["protein_name"])
psap_prediction["PSAP_score"] = clf.predict_proba(X)[:, 1]
psap_prediction["llps"] = y.values
psap_prediction["rank"] = 0
rank = psap_prediction.loc[psap_prediction["llps"] == 0,
"PSAP_score"].rank(ascending=False)
psap_prediction["rank"] = rank
# # Make directory for output
out_dir = Path(out_dir)
out_dir.mkdir(parents=True, exist_ok=True)
psap_prediction.to_csv(out_dir / f"prediction_{prefix}.csv")
# sklearn rf-model, serialize as json
import sklearn_json as skljson
from sklearn.ensemble import RandomForestClassifier
from sklearn.datasets import load_wine
from sklearn.model_selection import train_test_split
## load data
wine = load_wine()
## split train/tmp3.json
Xtrain, Xtest, Ytrain, Ytest = train_test_split(wine.data,
wine.target,
test_size=0.3)
model = RandomForestClassifier(n_estimators=10, max_depth=5, random_state=0)\
.fit(Xtrain, Ytrain)
## to json
skljson.to_json(model, "rf-model")
## from_json
model2 = skljson.from_json("rf-model")
score = model2.score(Xtest, Ytest)
print(score)
import sklearn_json as skljson
import pickle
deserialized_model = skljson.from_json('lr_model.json')
from sklearn.preprocessing import MinMaxScaler
with open('scaler.json', 'rb') as f:
scaler = pickle.load(f)
deserialized_model.predict(scaler.transform([[152, 6.5, 8.5, 0.72]]))
explainer = shap.TreeExplainer(clf)
shap_values = explainer.shap_values(X_train)
vals = np.abs(shap_values[0]).mean(0)
shap_importance = pd.Series(vals, index=X_train.columns).rename('shap')
shap_importance.sort_values(ascending=False, inplace=True)
imp = pd.concat([imp, shap_importance], axis=1)
imp.to_csv('feature_importnace.csv')
# Save model
import sklearn_json as skljson
file_name = os.path.join('models', 'random_forest.json')
clf_ser = serialize_random_forest(clf)
with open(file_name, 'w') as model_json:
json.dump(clf_ser, model_json)
deserialized_model = skljson.from_json(file_name)
deserialized_model.predict(X_test)
# add to dropbox
# import dropbox
# dbx = dropbox.Dropbox('RPqFmEm0LbUAAAAAAAAAAZ5Q4ZbVET-HQh18ixMUp6Gcx5lc0vMYMzMA2rueMjO6')
# with open(file_name, 'rb') as f:
# dbx.files_upload(f.read(), '/trend_labeling/' + file_name, mute = True)
### REFIT THE MODEL WITH MOST IMPORTANT FEATURES
# fi_cols = shap_values['col_name'].head(keep_important_features)
# X_train_important = X_train[fi_cols]
# X_test_important = X_test[fi_cols]
# clf = RandomForestClassifier(criterion='entropy',
# max_features=keep_important_features,
# min_weight_fraction_leaf=min_weight_fraction_leaf,
#pip install sklearn-json
import sklearn_json as skljson
file_name = "abc.json"
skljson.to_json(model, file_name)
deserialized_model = skljson.from_json('abc.json')
# read and save model in file
import json
import sklearn_json as skljson
from sklearn.model_selection import train_test_split
from sklearn.datasets import load_wine
f = json.load(open("tree_model", mode='r', encoding='utf-8'))
# print(f, type(f))
# print(f['meta'])
# ======================================
# 读取json文件的model，使用model
# ======================================
wine = load_wine()
Xtrain, Xtest, Ytrain, Ytest = train_test_split(wine.data,
wine.target,
test_size=0.3)
model = skljson.from_json("tree_model")
print(model.score(Xtrain, Ytrain))
print(model.score(Xtest, Ytest))
print(model.predict(Xtest))
print(Ytest)
