def test_save_and_load():
test_name = 'test-model-state'
model = Model()
model.fit(save=False)
path = model.save(model_name=test_name, upload=False)
model = Model.load(model_path=path)
assert len(model.predict(data.head(10))) > 0
os.remove(path)
def __init__(self):
self.show_image = False
self.show_border = False
self.show_overlay = False
self.save_full_image = False
self.output = False
self.start_x, self.start_y, self.w, self.h = 0, 0, 0, 0
self.keys_model = Model.load("ml/keys/keys_model.pkl")
def train_model():
model = Model()
model.fit(save=True, upload=True)
def test_upload_and_download():
test_name = 'test-model'
session = boto3.Session()
s3 = session.resource('s3')
obj = s3.Object(config.models_bucket, test_name)
obj.delete()
model = Model()
model.fit(save=False)
path = model.save(test_name, upload=False)
model.upload(test_name, model_path=path)
path = Model.download(model_name=test_name)
model = Model.load(model_path=path)
predictions = model.predict(data.head(10))
assert len(predictions) == 10
assert 1 in predictions
assert 0 in predictions
obj = s3.Object(config.models_bucket, test_name)
obj.delete()
model = Model()
model.fit(save=False)
model.save(model_name=test_name)
model = Model.load(model_name=test_name)
assert len(model.predict(data.head(10))) > 0
def test_train():
model = Model()
assert model.version == -1
model.fit(save=False)
assert model.version == 0
model.predict(data.head(10))
def test_get_data():
model = Model()
df = model.get_data()
assert len(df) > 0
def test_get_estimator():
model = Model()
estimator = model.get_estimator()
assert estimator is not None
import os
import boto3
import config
from ml.model import Model
data = Model.get_data()
def test_get_estimator():
model = Model()
estimator = model.get_estimator()
assert estimator is not None
def test_get_data():
model = Model()
df = model.get_data()
assert len(df) > 0
def test_train():
model = Model()
assert model.version == -1
model.fit(save=False)
assert model.version == 0
model.predict(data.head(10))
def test_save_and_load():
import pandas as pd
from flask import abort, Response, request, make_response, Flask
import config
from ml.model import Model
from ml.utils import get_version
logging.basicConfig()
logger = logging.getLogger(__name__)
logger.setLevel(logging.DEBUG)
app = Flask(__name__)
model_name = config.model_name
version = get_version(model_name)
#TODO cache
model = Model.load(model_name=model_name, version=version)
@app.route('/', methods=['POST'])
def live(event=None, context=None):
return Response('live')
def response(replay, status=200, json_dumps=True, mimetype='application/json'):
if json_dumps:
replay = json.dumps(replay)
return Response(response=replay, status=status, mimetype=mimetype)
def bad_request(message="", code=400):
logger.debug("bad request: %s" % message)