def set_data(self, data):
self.closeContext()
self.data = data
self.cb_pa.setEnabled(not isinstance(data, SqlTable))
self.cb_pc.setEnabled(not isinstance(data, SqlTable))
self.remove_all_rows()
self.add_button.setDisabled(data is None)
self.add_all_button.setDisabled(
data is None
or len(data.domain.variables) + len(data.domain.metas) > 100)
if not data:
self.data_desc = None
self.commit()
return
self.data_desc = report.describe_data_brief(data)
self.conditions = []
try:
self.openContext(data)
except Exception:
pass
variables = list(
filter_visible(chain(data.domain.variables, data.domain.metas)))
varnames = [v.name for v in variables]
if self.conditions:
for attr, cond_type, cond_value in self.conditions:
if attr in varnames:
self.add_row(varnames.index(attr), cond_type, cond_value)
elif variables:
self.add_row()
self.update_info(data, self.data_in_variables, "In: ")
self.unconditional_commit()
def set_data(self, data):
self.closeContext()
self.data = data
self.cb_pa.setEnabled(not isinstance(data, SqlTable))
self.cb_pc.setEnabled(not isinstance(data, SqlTable))
self.remove_all_rows()
self.add_button.setDisabled(data is None)
self.add_all_button.setDisabled(
data is None or
len(data.domain.variables) + len(data.domain.metas) > 100)
if not data:
self.data_desc = None
self.commit()
return
self.data_desc = report.describe_data_brief(data)
self.conditions = []
try:
self.openContext(data)
except Exception:
pass
if not self.conditions and len(data.domain.variables):
self.add_row()
self.update_info(data, self.data_in_variables)
for attr, cond_type, cond_value in self.conditions:
attrs = [a.name for a in
data.domain.variables + data.domain.metas]
if attr in attrs:
self.add_row(attrs.index(attr), cond_type, cond_value)
self.unconditional_commit()
def set_data(self, data):
self.closeContext()
self.data = data
self.cb_pa.setEnabled(not isinstance(data, SqlTable))
self.cb_pc.setEnabled(not isinstance(data, SqlTable))
self.remove_all_rows()
self.add_button.setDisabled(data is None)
self.add_all_button.setDisabled(
data is None or
len(data.domain.variables) + len(data.domain.metas) > 100)
if not data:
self.data_desc = None
self.commit()
return
self.data_desc = report.describe_data_brief(data)
self.conditions = []
try:
self.openContext(data)
except Exception:
pass
variables = list(self._visible_variables(self.data.domain))
varnames = [v.name for v in variables]
if self.conditions:
for attr, cond_type, cond_value in self.conditions:
if attr in varnames:
self.add_row(varnames.index(attr), cond_type, cond_value)
elif variables:
self.add_row()
self.update_info(data, self.data_in_variables, "In: ")
self.unconditional_commit()
def set_data(self, data):
self.closeContext()
self.data = data
self.cb_pa.setEnabled(not isinstance(data, SqlTable))
self.cb_pc.setEnabled(not isinstance(data, SqlTable))
self.remove_all_rows()
self.add_button.setDisabled(data is None)
self.add_all_button.setDisabled(
data is None
or len(data.domain.variables) + len(data.domain.metas) > 100)
if not data:
self.data_desc = None
self.commit()
return
self.data_desc = report.describe_data_brief(data)
self.conditions = []
try:
self.openContext(data)
except Exception:
pass
if not self.conditions and len(data.domain.variables):
self.add_row()
self.update_info(data, self.data_in_variables, "In: ")
for attr, cond_type, cond_value in self.conditions:
attrs = [a.name for a in data.domain.variables + data.domain.metas]
if attr in attrs:
self.add_row(attrs.index(attr), cond_type, cond_value)
self.unconditional_commit()
def commit(self):
matching_output = self.data
non_matching_output = None
self.error()
if self.data:
domain = self.data.domain
conditions = []
for attr_name, oper_idx, values in self.conditions:
attr_index = domain.index(attr_name)
attr = domain[attr_index]
operators = self.Operators[type(attr)]
opertype, _ = operators[oper_idx]
if attr.is_continuous:
if any(not v for v in values):
continue
# Parse datetime strings into floats
if isinstance(attr, TimeVariable):
try:
values = [attr.parse(v) for v in values]
except ValueError as e:
self.error(e.args[0])
return
filter = data_filter.FilterContinuous(
attr_index, opertype, *[float(v) for v in values])
elif attr.is_string:
filter = data_filter.FilterString(
attr_index, opertype, *[str(v) for v in values])
else:
if opertype == FilterDiscreteType.IsDefined:
f_values = None
else:
if not values or not values[0]:
continue
values = [attr.values[i - 1] for i in values]
if opertype == FilterDiscreteType.Equal:
f_values = {values[0]}
elif opertype == FilterDiscreteType.NotEqual:
f_values = set(attr.values)
f_values.remove(values[0])
elif opertype == FilterDiscreteType.In:
f_values = set(values)
else:
raise ValueError("invalid operand")
filter = data_filter.FilterDiscrete(attr_index, f_values)
conditions.append(filter)
if conditions:
self.filters = data_filter.Values(conditions)
matching_output = self.filters(self.data)
self.filters.negate = True
non_matching_output = self.filters(self.data)
# if hasattr(self.data, "name"):
# matching_output.name = self.data.name
# non_matching_output.name = self.data.name
purge_attrs = self.purge_attributes
purge_classes = self.purge_classes
if (purge_attrs or purge_classes) and \
not isinstance(self.data, SqlTable):
attr_flags = sum([
Remove.RemoveConstant * purge_attrs,
Remove.RemoveUnusedValues * purge_attrs
])
class_flags = sum([
Remove.RemoveConstant * purge_classes,
Remove.RemoveUnusedValues * purge_classes
])
# same settings used for attributes and meta features
remover = Remove(attr_flags, class_flags, attr_flags)
matching_output = remover(matching_output)
non_matching_output = remover(non_matching_output)
self.send("Matching Data", matching_output)
self.send("Unmatched Data", non_matching_output)
self.match_desc = report.describe_data_brief(matching_output)
self.nonmatch_desc = report.describe_data_brief(non_matching_output)
self.update_info(matching_output, self.data_out_rows, "Out: ")
def commit(self):
matching_output = self.data
non_matching_output = None
if self.data:
domain = self.data.domain
conditions = []
for attr_name, oper, values in self.conditions:
attr_index = domain.index(attr_name)
attr = domain[attr_index]
if attr.is_continuous:
if any(not v for v in values):
continue
filter = data_filter.FilterContinuous(
attr_index, oper, *[float(v) for v in values])
elif attr.is_string:
filter = data_filter.FilterString(
attr_index, oper, *[str(v) for v in values])
else:
if oper == 3:
f_values = None
else:
if not values or not values[0]:
continue
values = [attr.values[i-1] for i in values]
if oper == 0:
f_values = {values[0]}
elif oper == 1:
f_values = set(attr.values)
f_values.remove(values[0])
elif oper == 2:
f_values = set(values)
else:
raise ValueError("invalid operand")
filter = data_filter.FilterDiscrete(attr_index, f_values)
conditions.append(filter)
if conditions:
filters = data_filter.Values(conditions)
matching_output = filters(self.data)
filters.negate = True
non_matching_output = filters(self.data)
# if hasattr(self.data, "name"):
# matching_output.name = self.data.name
# non_matching_output.name = self.data.name
purge_attrs = self.purge_attributes
purge_classes = self.purge_classes
if (purge_attrs or purge_classes) and \
not isinstance(self.data, SqlTable):
attr_flags = sum([Remove.RemoveConstant * purge_attrs,
Remove.RemoveUnusedValues * purge_attrs])
class_flags = sum([Remove.RemoveConstant * purge_classes,
Remove.RemoveUnusedValues * purge_classes])
# same settings used for attributes and meta features
remover = Remove(attr_flags, class_flags, attr_flags)
matching_output = remover(matching_output)
non_matching_output = remover(non_matching_output)
self.send("Matching Data", matching_output)
self.send("Unmatched Data", non_matching_output)
self.match_desc = report.describe_data_brief(matching_output)
self.nonmatch_desc = report.describe_data_brief(non_matching_output)
self.update_info(matching_output, self.data_out_rows)
def commit(self):
matching_output = self.data
non_matching_output = None
annotated_output = None
self.Error.clear()
if self.data:
domain = self.data.domain
conditions = []
for attr_name, oper_idx, values in self.conditions:
attr_index = domain.index(attr_name)
attr = domain[attr_index]
operators = self.Operators[type(attr)]
opertype, _ = operators[oper_idx]
if attr.is_continuous:
try:
floats = self._values_to_floats(attr, values)
except ValueError as e:
self.Error.parsing_error(e.args[0])
return
if floats is None:
continue
filter = data_filter.FilterContinuous(
attr_index, opertype, *floats)
elif attr.is_string:
filter = data_filter.FilterString(
attr_index, opertype, *[str(v) for v in values])
else:
if opertype == FilterDiscreteType.IsDefined:
f_values = None
else:
if not values or not values[0]:
continue
values = [attr.values[i - 1] for i in values]
if opertype == FilterDiscreteType.Equal:
f_values = {values[0]}
elif opertype == FilterDiscreteType.NotEqual:
f_values = set(attr.values)
f_values.remove(values[0])
elif opertype == FilterDiscreteType.In:
f_values = set(values)
else:
raise ValueError("invalid operand")
filter = data_filter.FilterDiscrete(attr_index, f_values)
conditions.append(filter)
if conditions:
self.filters = data_filter.Values(conditions)
matching_output = self.filters(self.data)
self.filters.negate = True
non_matching_output = self.filters(self.data)
row_sel = np.in1d(self.data.ids, matching_output.ids)
annotated_output = create_annotated_table(self.data, row_sel)
# if hasattr(self.data, "name"):
# matching_output.name = self.data.name
# non_matching_output.name = self.data.name
purge_attrs = self.purge_attributes
purge_classes = self.purge_classes
if (purge_attrs or purge_classes) and \
not isinstance(self.data, SqlTable):
attr_flags = sum([
Remove.RemoveConstant * purge_attrs,
Remove.RemoveUnusedValues * purge_attrs
])
class_flags = sum([
Remove.RemoveConstant * purge_classes,
Remove.RemoveUnusedValues * purge_classes
])
# same settings used for attributes and meta features
remover = Remove(attr_flags, class_flags, attr_flags)
matching_output = remover(matching_output)
non_matching_output = remover(non_matching_output)
annotated_output = remover(annotated_output)
if matching_output is not None and not len(matching_output):
matching_output = None
if non_matching_output is not None and not len(non_matching_output):
non_matching_output = None
if annotated_output is not None and not len(annotated_output):
annotated_output = None
self.Outputs.matching_data.send(matching_output)
self.Outputs.unmatched_data.send(non_matching_output)
self.Outputs.annotated_data.send(annotated_output)
self.match_desc = report.describe_data_brief(matching_output)
self.nonmatch_desc = report.describe_data_brief(non_matching_output)
self.update_info(matching_output, self.data_out_rows, "Out: ")
def commit(self):
matching_output = self.data
non_matching_output = None
self.Error.clear()
if self.data:
domain = self.data.domain
conditions = []
for attr_name, oper_idx, values in self.conditions:
attr_index = domain.index(attr_name)
attr = domain[attr_index]
operators = self.Operators[type(attr)]
opertype, _ = operators[oper_idx]
if attr.is_continuous:
try:
floats = self._values_to_floats(attr, values)
except ValueError as e:
self.Error.parsing_error(e.args[0])
return
if floats is None:
continue
filter = data_filter.FilterContinuous(
attr_index, opertype, *floats)
elif attr.is_string:
filter = data_filter.FilterString(
attr_index, opertype, *[str(v) for v in values])
else:
if opertype == FilterDiscreteType.IsDefined:
f_values = None
else:
if not values or not values[0]:
continue
values = [attr.values[i-1] for i in values]
if opertype == FilterDiscreteType.Equal:
f_values = {values[0]}
elif opertype == FilterDiscreteType.NotEqual:
f_values = set(attr.values)
f_values.remove(values[0])
elif opertype == FilterDiscreteType.In:
f_values = set(values)
else:
raise ValueError("invalid operand")
filter = data_filter.FilterDiscrete(attr_index, f_values)
conditions.append(filter)
if conditions:
self.filters = data_filter.Values(conditions)
matching_output = self.filters(self.data)
self.filters.negate = True
non_matching_output = self.filters(self.data)
# if hasattr(self.data, "name"):
# matching_output.name = self.data.name
# non_matching_output.name = self.data.name
purge_attrs = self.purge_attributes
purge_classes = self.purge_classes
if (purge_attrs or purge_classes) and \
not isinstance(self.data, SqlTable):
attr_flags = sum([Remove.RemoveConstant * purge_attrs,
Remove.RemoveUnusedValues * purge_attrs])
class_flags = sum([Remove.RemoveConstant * purge_classes,
Remove.RemoveUnusedValues * purge_classes])
# same settings used for attributes and meta features
remover = Remove(attr_flags, class_flags, attr_flags)
matching_output = remover(matching_output)
non_matching_output = remover(non_matching_output)
self.Outputs.matching_data.send(matching_output)
self.Outputs.unmatched_data.send(non_matching_output)
self.match_desc = report.describe_data_brief(matching_output)
self.nonmatch_desc = report.describe_data_brief(non_matching_output)
self.update_info(matching_output, self.data_out_rows, "Out: ")
def commit(self):
matching_output = self.data
non_matching_output = None
if self.data:
domain = self.data.domain
conditions = []
for attr_name, oper, values in self.conditions:
attr_index = domain.index(attr_name)
attr = domain[attr_index]
if attr.is_continuous:
if any(not v for v in values):
continue
filter = data_filter.FilterContinuous(
attr_index, oper, *[float(v) for v in values])
elif attr.is_string:
filter = data_filter.FilterString(
attr_index, oper, *[str(v) for v in values])
else:
if oper == 3:
f_values = None
else:
if not values or not values[0]:
continue
values = [attr.values[i - 1] for i in values]
if oper == 0:
f_values = {values[0]}
elif oper == 1:
f_values = set(attr.values)
f_values.remove(values[0])
elif oper == 2:
f_values = set(values)
else:
raise ValueError("invalid operand")
filter = data_filter.FilterDiscrete(attr_index, f_values)
conditions.append(filter)
if conditions:
filters = data_filter.Values(conditions)
matching_output = filters(self.data)
filters.negate = True
non_matching_output = filters(self.data)
# if hasattr(self.data, "name"):
# matching_output.name = self.data.name
# non_matching_output.name = self.data.name
purge_attrs = self.purge_attributes
purge_classes = self.purge_classes
if (purge_attrs or purge_classes) and \
not isinstance(self.data, SqlTable):
attr_flags = sum([
Remove.RemoveConstant * purge_attrs,
Remove.RemoveUnusedValues * purge_attrs
])
class_flags = sum([
Remove.RemoveConstant * purge_classes,
Remove.RemoveUnusedValues * purge_classes
])
# same settings used for attributes and meta features
remover = Remove(attr_flags, class_flags, attr_flags)
matching_output = remover(matching_output)
non_matching_output = remover(non_matching_output)
self.send("Matching Data", matching_output)
self.send("Unmatched Data", non_matching_output)
self.match_desc = report.describe_data_brief(matching_output)
self.nonmatch_desc = report.describe_data_brief(non_matching_output)
self.update_info(matching_output, self.data_out_rows)
