def rand_param(parameters):
"""
Obtains the parameters for the random dataset generator.
:param parameters: The parameters for the creation (number of rows, numbers of columns,...).
:return: A tuple of the parameters.
"""
num_rows, num_col, values = 0, 0, []
if parameters["columns"]:
num_col = int(parameters["columns"])
else:
print('How many columns?')
al.voice('How many columns?')
query = al.query_input()
while not query.isnumeric():
print(
'Incorrect input.\nIt is not a number.\nPlease introduce one.')
al.voice(
'Incorrect input.\nIt is not a number.\nPlease introduce one.')
query = al.query_input()
num_col = int(query)
if parameters["rows"]:
num_rows = int(parameters["rows"])
else:
print('How many rows?')
al.voice('How many rows?')
query = al.query_input()
while not query.isnumeric():
print(
'Incorrect input.\nIt is not a number.\nPlease introduce one.')
al.voice(
'Incorrect input.\nIt is not a number.\nPlease introduce one.')
query = al.query_input()
num_rows = int(query)
if parameters["values"]:
values = parameters["values"]
else:
print('What is the minimum value?')
al.voice('What is the minimum value?')
query = al.query_input()
while not al.isnumber(query):
print(
'Incorrect input.\nIt is not a number.\nPlease introduce one:')
al.voice(
'Incorrect input.\nIt is not a number.\nPlease introduce one.')
query = al.query_input()
values.append(float(query))
print('And the maximum?')
al.voice('And the maximum?')
query = al.query_input()
while not al.isnumber(query):
print(
'Incorrect input.\nIt is not a number.\nPlease introduce one:')
al.voice(
'Incorrect input.\nIt is not a number.\nPlease introduce one.')
query = al.query_input()
values.append(float(query))
return num_rows, num_col, values
def get_subdataset_rows(parameters):
"""
Obtains a subset of the dataset by its rows.
:param parameters: The parameter of the function(dataset name,...).
"""
workspace = al.Workspace()
data_name = parameters['Dataset']
dataset = workspace.get_dataset(data_name)
if parameters["from"]:
index_a = int(parameters["from"])
else:
print('From what row number?')
al.voice('From what row number?')
query = al.query_input()
while not al.isnumber(query):
print(
'Incorrect input.\nIt is not a number.\nPlease introduce one:')
al.voice(
'Incorrect input.\nIt is not a number.\nPlease introduce one.')
query = al.query_input()
index_a = int(query)
if parameters["to"]:
index_b = int(parameters['to'])
else:
print('To what row number?')
al.voice('To what row number?')
query = al.query_input()
while not al.isnumber(query):
print(
'Incorrect input.\nIt is not a number.\nPlease introduce one:')
al.voice(
'Incorrect input.\nIt is not a number.\nPlease introduce one.')
query = al.query_input()
index_b = int(query)
if index_b < index_a:
print(
'This operation cannot be done.\nThe starting row number is greater than the last row number.'
)
raise Exception()
dataset = dataset.iloc[index_a:index_b]
num = workspace.get_counter('sub')
name = 'subrow' + str(num) + data_name
workspace.save_dataset(name, dataset)
txt = 'The sub-dataset by the rows is saved as ' + name
print(txt)
def split_by_rows(parameters):
"""
Split a dataset into n datasets of m rows.
:param parameters: The parameters of the function (dataset name, size of the split dataset for the rows).
"""
workspace = al.Workspace()
name_data = parameters['Dataset']
dataset = workspace.get_dataset(name_data)
if parameters['split']:
div = int(parameters['split'])
else:
print('How many rows will each dataset have?')
query = al.query_input()
while not al.isnumber(query):
print(
'Incorrect input.\nIt is not a number.\nPlease introduce one:')
query = al.query_input()
div = int(query)
it = 0
names = []
while it < dataset.index.size:
div_dataset = dataset.iloc[it:it + div]
num = workspace.get_counter('split')
name = name_data + 'r' + str(num)
names.append(name)
workspace.save_dataset(name, div_dataset)
it = it + div
print('The splits of ' + name_data + ' are saved as: ' + str(names)[1:-1])