def setUpClass(cls):
super(Test, cls).setUpClass()
dbPath = "testDb.sqlite"
if os.path.exists(dbPath):
os.remove(dbPath)
store = Store(dbPath)
assert isinstance(store, Store)
for k1 in range(2**3):
buildParameter = BuildParameter(label = "test" + str(k1))
agent = Agent()
for epoch in range(2**4):
agentMemento = agent.createMemento()
buildParameterMemento = buildParameter.createMemento()
buildParameterKey = buildParameter.key
buildParameterLabel = buildParameter.label
storeField = StoreField(agentMemento, epoch, buildParameterMemento, buildParameterKey, buildParameterLabel)
assert isinstance(storeField, StoreField)
store.append(storeField)
cls.dbPath = dbPath
def test002(self):
store = Store(self.dbPath)
assert isinstance(store, Store)
for k1 in range(2**3):
buildParameter = BuildParameter(label="test" + str(k1))
agent = Agent()
for epoch in range(2**4):
agentMemento = agent.createMemento()
buildParameterMemento = buildParameter.createMemento()
buildParameterKey = buildParameter.key
buildParameterLabel = buildParameter.label
storeField = StoreField(agentMemento, epoch,
buildParameterMemento,
buildParameterKey, buildParameterLabel)
assert isinstance(storeField, StoreField)
store.append(storeField)
store.update_db()
for storeField in store.restore("test%", 1):
assert isinstance(storeField, StoreField)
class Manager(object):
def __init__(self, pcid, cid, datamonth):
self.pcid = pcid
self.cid = cid
self.datamonth = datamonth
self.save = Store(pcid, cid)
@decorator.process
def make_mat(self):
"""
Select the useful targets and submarkets
Organize raw data
Embedding the scores and make the matrix
Construct the relationship between line No. and model
PCA dimension reduction
:return:
"""
info_target = get_targets_info(self.pcid,
self.cid,
self.datamonth,
src="LOCAL")
targets = select_targets(info_target, store=self.save)
urls = get_urls(self.pcid, self.cid, self.datamonth, src="LOCAL")
_ = map_model_urls(urls, self.save)
info_submarket = get_submarket_info(self.pcid,
self.cid,
self.datamonth,
src="LOCAL")
_, sm_hot, = select_submarkets(info_submarket, store=self.save)
info = reconstruct_info(info_target, info_submarket, targets, sm_hot)
mat, map_line = embedding(info, targets, self.save)
pca_mat, _ = pca(mat, self.save)
# mat, map_line = embedding(info, targets, self.save, normal="NORMAL")
# pca_mat = mat
return pca_mat, map_line, info, sm_hot
@decorator.process
def describe_submarket(self, mat, map_line, info, sm_hot):
"""
Describe submarkets as vector
:param mat:
:param map_line:
:param info:
:param sm_hot:
:return:
"""
models = select_top_model(info, sm_hot, self.save)
lines = model_map_to_line(models, map_line)
portraits_sm = establish_portraits_sm(lines, mat, self.save)
# portraits_sm = establish_portraits_sm_by_lda(lines, mat, self.save)
return portraits_sm
@decorator.process
def position_models(self, mat, map_line, portraits_sm):
"""
Calculate distance between every model and every submarket
:param mat:
:param map_line:
:param portraits_sm:
:return:
"""
distances, similarity = find_position(mat, map_line, portraits_sm,
self.save)
return distances, similarity
@decorator.process
def append_feature(self, mat, map_line):
sale_info = get_sale_info(self.pcid,
self.cid,
self.datamonth,
src="LOCAL")
sale_mat = map_feature_info(sale_info, map_line, self.save, "sale")
mat = append_method(mat, sale_mat)
sku_info = get_sku_info(self.pcid,
self.cid,
self.datamonth,
src="LOCAL")
# getting sku from table product_brain need loading json
sku_info = load_sku_json(sku_info)
sku_mat = map_feature_info(sku_info, map_line, self.save, "sku")
mat = append_method(mat, sku_mat)
self.save.append("mat_plus",
data=mat,
layout="np.array",
val_type="float")
return mat
@decorator.process
def run(self):
"""
Process the tasks and save datum and results
:return:
"""
mat, map_line, info, sm_hot = self.make_mat()
mat = self.append_feature(mat, map_line)
portraits_sm = self.describe_submarket(mat, map_line, info, sm_hot)
_, _ = self.position_models(mat, map_line, portraits_sm)
self.save.save()
@decorator.process
def query(self, method="PCA"):
"""
Query the distance and similarity between model and hot submarkets
: demo: brand 1号宝贝 model OUYOU0011
: demo: brand fronton/弗朗顿 model CB-666D
:return:
"""
if "LDA" == method:
query_position_lda(self.save)
else:
query_position(self.save)
def main(self):
choice = input(
"\ninput 0 to generate result，input 1 to query result:\n")
if "0" == str(choice):
self.run()
else:
method = input("\ninput suffix:\n")
self.query(method)
