def test_dense_output_format(self):
reader = DataIO()
reader.set_tracker(self.tracker)
component_params = {
"DataIOParam": {
"input_format": "dense"
},
"role": {
"guest": [9999],
"host": [10000],
"arbiter": [10000]
},
"local": {
"role": "guest",
"party_id": 9999
}
}
reader.run(component_params, self.args1)
data = reader.save_data().collect()
result = dict(data)
self.assertTrue(type(result['a']).__name__ == "Instance")
self.assertTrue(type(result['b']).__name__ == "Instance")
vala = result['a']
features = vala.features
weight = vala.weight
label = vala.label
self.assertTrue(np.abs(weight - 1.0) < consts.FLOAT_ZERO)
self.assertTrue(type(features).__name__ == "ndarray")
self.assertTrue(label == None)
self.assertTrue(features.shape[0] == 6)
self.assertTrue(features.dtype == "float64")
def test_sparse_output_format(self):
reader = DataIO()
reader.set_tracker(self.tracker)
component_params = {
"DataIOParam": {
"output_format": "sparse",
"input_format": "dense"
},
"role": {
"guest": [9999],
"host": [10000],
"arbiter": [10000]
},
"local": {
"role": "guest",
"party_id": 9999
}
}
reader.run(component_params, self.args1)
data = reader.save_data().collect()
result = dict(data)
vala = result['a']
features = vala.features
self.assertTrue(type(features).__name__ == "SparseVector")
self.assertTrue(len(features.sparse_vec) == 4)
self.assertTrue(features.shape == 6)
def test_with_label(self):
reader = DataIO()
reader.set_tracker(self.tracker)
component_params = {
"DataIOParam": {
"output_format": "dense",
"input_format": "dense",
"with_label": True,
"label_name": "x3"
},
"role": {
"guest": [9999],
"host": [10000],
"arbiter": [10000]
},
"local": {
"role": "guest",
"party_id": 9999
}
}
reader.run(component_params, self.args1)
data = reader.save_data().collect()
result = dict(data)
vala = result['a']
label = vala.label
features = vala.features
self.assertTrue(label == -1)
self.assertTrue(features.shape[0] == 5)
def test_sparse_output_format(self):
reader = DataIO()
reader.set_tracker(self.tracker)
component_params = {
"DataIOParam": {
"output_format": "sparse",
"input_format": "sparse",
"delimitor": ' ',
"defualt_value": 2**30
},
"role": {
"guest": [9999],
"host": [10000],
"arbiter": [10000]
},
"local": {
"role": "guest",
"party_id": 9999
}
}
reader.run(component_params, self.args)
data = reader.save_data().collect()
for i in range(100):
self.assertTrue(
insts[i][1].features.get_shape() == self.max_feature + 1)
self.assertTrue(insts[i][1].label == i % 2)
original_feat = {}
row = self.data[i][1].split(" ")
for j in range(1, len(row)):
fid, val = row[j].split(":", -1)
original_feat[int(fid)] = float(val)
self.assertTrue(original_feat == insts[i][1].features.sparse_vec)
def test_sparse_output_format(self):
dataio = DataIO()
dataio.set_tracker(TrackerMock())
component_params = {
"DataIOParam": {
"output_format": "sparse",
"input_format": "sparse",
"delimitor": ' '
},
"role": {
"guest": [9999],
"host": [10000],
"arbiter": [10000]
},
"local": {
"role": "guest",
"party_id": 9999
}
}
dataio.run(component_params, self.args)
insts = list(dataio.save_data().collect())
for i in range(100):
features = insts[i][1].features
self.assertTrue(type(features).__name__ == "SparseVector")
self.assertTrue(features.get_shape() == self.max_feature + 1)
self.assertTrue(insts[i][1].label == i % 2)
row = self.data[i][1].split(" ")
for j in range(1, len(row)):
fid, val = row[j].split(":", -1)
self.assertTrue(
np.fabs(features.get_data(int(fid)) -
float(val)) < consts.FLOAT_ZERO)
def test_missing_value_fill(self):
reader = DataIO()
reader.set_tracker(self.tracker)
component_params = {
"DataIOParam": {
"output_format": "sparse",
"input_format": "dense",
"default_value": 100,
"with_label": False,
"missing_fill": True,
"missing_fill_method": "designated",
"data_type": "int"
},
"role": {
"guest": [9999],
"host": [10000],
"arbiter": [10000]
},
"local": {
"role": "guest",
"party_id": 9999
}
}
reader.run(component_params, self.args2)
data = reader.save_data().collect()
result = dict(data)
features = result['a'].features
for i in range(1, 5):
self.assertTrue(features.get_data(i) == 100)
def test_dense_output_format(self):
dataio = DataIO()
dataio.set_tracker(TrackerMock())
component_params = {
"output_format": "dense",
"input_format": "sparse",
"delimitor": ' '
}
cpn_input = get_cpn_input(self.dataset, component_params)
dataio.run(cpn_input)
insts = list(dataio.save_data().collect())
for i in range(100):
features = insts[i][1].features
self.assertTrue(type(features).__name__ == "ndarray")
self.assertTrue(features.shape[0] == self.max_feature + 1)
self.assertTrue(insts[i][1].label == i % 2)
row = self.data[i][1].split(" ")
ori_feat = [0 for i in range(self.max_feature + 1)]
for j in range(1, len(row)):
fid, val = row[j].split(":", -1)
ori_feat[int(fid)] = float(val)
ori_feat = np.asarray(ori_feat, dtype="float64")
self.assertTrue(
np.abs(ori_feat - features).any() < consts.FLOAT_ZERO)
def test_tag_with_value_dense_output_format(self):
dataio = DataIO()
dataio.set_tracker(TrackerMock())
component_params = {
"DataIOParam": {
"output_format": "dense",
"input_format": "tag",
"delimitor": ' ',
"data_type": "float",
"with_label": False,
"tag_with_value": True
},
"role": {
"guest": [9999],
"host": [10000],
"arbiter": [10000]
},
"local": {
"role": "guest",
"party_id": 9999
}
}
dataio.run(component_params, self.args2)
tag_insts = dataio.save_data()
features = [inst.features for key, inst in tag_insts.collect()]
tags = set()
for row in self.data_with_value:
tag_list = []
for tag_with_value in row[1].split(" ", -1):
tag_list.append(tag_with_value.split(":")[0])
tags |= set(tag_list)
tags = sorted(tags)
tag_dict = dict(zip(tags, range(len(tags))))
for i in range(len(self.data_with_value)):
ori_feature = [0 for i in range(len(tags))]
for tag_with_value in self.data_with_value[i][1].split(" ", -1):
tag = tag_with_value.split(":", -1)[0]
val = float(tag_with_value.split(":", -1)[1])
ori_feature[tag_dict.get(tag)] = val
ori_feature = np.asarray(ori_feature, dtype='float64')
self.assertTrue(
np.abs(ori_feature - features).all() < consts.FLOAT_ZERO)
def test_sparse_output_format(self):
reader = DataIO()
reader.set_tracker(self.tracker)
component_params = {"DataIOParam":
{"output_format": "sparse",
"input_format": "dense"
}
}
reader.run(component_params, self.args1)
data = reader.save_data().collect()
result = dict(data)
vala = result['a']
features = vala.features
self.assertTrue(type(features).__name__ == "SparseVector")
self.assertTrue(len(features.sparse_vec) == 4)
self.assertTrue(features.shape == 6)
def test_with_label(self):
reader = DataIO()
reader.set_tracker(self.tracker)
component_params = {
"DataIOParam": {
"output_format": "dense",
"input_format": "dense",
"with_label": True,
"label_idx": 2
}
}
reader.run(component_params, self.args1)
data = reader.save_data().collect()
result = dict(data)
vala = result['a']
label = vala.label
features = vala.features
self.assertTrue(label == -1)
self.assertTrue(features.shape[0] == 5)
def test_tag_sparse_output_format(self):
dataio = DataIO()
dataio.set_tracker(TrackerMock())
component_params = {
"DataIOParam": {
"output_format": "sparse",
"input_format": "tag",
"delimitor": ' ',
"data_type": "int",
"with_label": False,
"tag_with_value": False
},
"role": {
"guest": [9999],
"host": [10000],
"arbiter": [10000]
},
"local": {
"role": "guest",
"party_id": 9999
}
}
dataio.run(component_params, self.args1)
tag_insts = dataio.save_data()
features = [inst.features for key, inst in tag_insts.collect()]
tags = set()
for row in self.data:
tags |= set(row[1].split(" ", -1))
tags = sorted(tags)
tag_dict = dict(zip(tags, range(len(tags))))
for i in range(len(self.data)):
ori_feature = {}
for tag in self.data[i][1].split(" ", -1):
ori_feature[tag_dict.get(tag)] = 1
self.assertTrue(ori_feature == features[i].sparse_vec)
def test_tag_with_value_sparse_output_format(self):
dataio = DataIO()
dataio.set_tracker(TrackerMock())
component_params = {
"output_format": "sparse",
"input_format": "tag",
"delimitor": ' ',
"data_type": "float",
"with_label": False,
"tag_with_value": True,
"tag_value_delimitor": ":"
}
cpn_input = get_cpn_input(self.dataset2, component_params)
dataio.run(cpn_input)
tag_insts = dataio.save_data()
features = [inst.features for key, inst in tag_insts.collect()]
tags = set()
for row in self.data_with_value:
tag_list = []
for tag_with_value in row[1].split(" ", -1):
tag_list.append(tag_with_value.split(":")[0])
tags |= set(tag_list)
tags = sorted(tags)
tag_dict = dict(zip(tags, range(len(tags))))
for i in range(len(self.data_with_value)):
ori_feature = {}
for tag_with_value in self.data_with_value[i][1].split(" ", -1):
idx = tag_dict.get(tag_with_value.split(":")[0])
val = float(tag_with_value.split(":")[1])
self.assertTrue(
np.abs(val -
features[i].get_data(idx)) < consts.FLOAT_ZERO)
def test_sparse_output_format(self):
dataio = DataIO()
dataio.set_tracker(TrackerMock())
component_params = {
"output_format": "sparse",
"input_format": "sparse",
"delimitor": ' '
}
cpn_input = get_cpn_input(self.dataset, component_params)
dataio.run(cpn_input)
insts = list(dataio.save_data().collect())
for i in range(100):
features = insts[i][1].features
self.assertTrue(type(features).__name__ == "SparseVector")
self.assertTrue(features.get_shape() == self.max_feature + 1)
self.assertTrue(insts[i][1].label == i % 2)
row = self.data[i][1].split(" ")
for j in range(1, len(row)):
fid, val = row[j].split(":", -1)
self.assertTrue(
np.fabs(features.get_data(int(fid)) -
float(val)) < consts.FLOAT_ZERO)