def BoostingStump(x, y, estimate_intercept=True, no_of_variables=None, *args, **kwargs):
'''
Boosting stump with Pandas variables
Input
--------
x Pandas DataFrame or Series of predictive variables
y Pandas DataFrame or Series of target variables
estimate_intercept Boolean
no_of_variables integer, the number of variables to use
Output
--------
ans Pandas DataFrame with predictive variable name, target variable name,
stump decision point of each x variable and weight corresponding to each x variable
Notes
--------
The target is to maximize the success rate of predicting the sign of y in-sample, weighted by the magnitude of y.
Use caution to normalize y before estimation.
x variables are used in the order that they are given.
If there are multiple columns in y, each column will be predicted independently.
'''
myx, myy = give_me_pandas_variables(x, y)
if no_of_variables is not None:
if np.size(myx, 1) > no_of_variables:
myx = myx.iloc[:, :no_of_variables]
ans = []
for i in xrange(np.size(myy, 1)):
try:
ans.append(estimate_boosting_stump(myx, myy.iloc[:, i], estimate_intercept))
except Exception as e:
logger.warning('Boosting stump failed at variable %s: %s' % (str(myy.columns[i]), str(e)))
return pd.concat(ans, axis=0)
def get_series_all_release(series_name, start_date=None, end_date=None):
read_script = get_release_read_script(RELEASE_TABLE_NAME, index_range=(start_date, end_date), column_list=[series_name])
success, data = du.execute_sql_output_script(DATABASE_NAME, read_script)
if success:
return pd.DataFrame(np.array(data), columns=['time_index', 'realtime_start', 'series_name', 'value']) if len(data) > 0 else None
else:
logger.warning('Failed to read data: ' + str(data))
def get_table_column_values(database_name,
table_name,
column_name=TIMESERIES_COLUMN_NAME):
read_script = get_table_column_value_scripts(table_name, column_name)
success, data = execute_sql_output_script(database_name, read_script)
if success:
return [x[0] for x in data] if len(data) > 0 else None
else:
logger.warning('Failed to read column names: ' + str(data))
def get_database_connection(database_name='mysql'):
try:
return mdb.connect(host=HOST,
user=USER,
passwd=PASSWORD,
db=database_name)
except Exception as e:
logger.warning('Failed to establish databse connection: ' + str(e))
return None
def release_bulk_insert(data, table_name, series_name):
delete_script = get_release_bulk_delete_script(data, table_name, series_name)
e = du.execute_sql_input_script(DATABASE_NAME, delete_script)
if e is not None:
logger.warning('Failed to clear data from table: ' + str(e))
else:
insert_script = get_release_bulk_insert_script(data, table_name, series_name)
e = du.execute_sql_input_script(DATABASE_NAME, insert_script)
if e is not None:
logger.warning('Failed to insert data: ' + str(e))
def pandas_delete(database_name,
table_name,
column_name,
index_name,
value_name,
index_range=None,
column_list=None,
data_name=None):
delete_script = get_pandas_delete_script(table_name, column_name,
index_name, value_name,
index_range, column_list,
data_name)
e = execute_sql_input_script(database_name, delete_script)
if e is not None:
logger.warning('Failed to delete data from table: ' + str(e))
def pandas_read(database_name,
table_name,
column_name,
index_name,
value_name,
index_range=None,
column_list=None,
data_name=None):
read_script = get_pandas_read_script(table_name, column_name, index_name,
value_name, index_range, column_list,
data_name)
success, data = execute_sql_output_script(database_name, read_script)
if success:
return get_pandas_output(data, column_name, index_name,
value_name) if len(data) > 0 else None
else:
logger.warning('Failed to read data: ' + str(data))
def pandas_bulk_insert(data,
database_name,
table_name,
column_name,
index_name,
value_name,
data_name=None,
data_column_name=None):
delete_script = get_pandas_bulk_delete_script(data, table_name,
column_name, index_name,
data_name, data_column_name)
e = execute_sql_input_script(database_name, delete_script)
if e is not None:
logger.warning('Failed to clear data from table: ' + str(e))
else:
insert_script = get_pandas_bulk_insert_script(data, table_name,
column_name, index_name,
value_name, data_name,
data_column_name)
e = execute_sql_input_script(database_name, insert_script)
if e is not None:
logger.warning('Failed to insert data: ' + str(e))
def create_table(database_name, table_name, table_format):
script = CREATE_TABLE_IF_NOT_EXISTS % (table_name, table_format)
e = execute_sql_input_script(database_name, script)
if e is not None:
logger.warning('Failed to create table: ' + str(e))
