import us
import pandas as pd
import numpy as np
from cytoolz.itertoolz import unique
from cytoolz.functoolz import thread_last, identity
from cytoolz.curried import map, filter, curry
from survey_stats import pdutil
# import sys
# import traceback as tb
from survey_stats import log
logger = log.getLogger(__name__)
US_STATES_FIPS_INTS = thread_last(us.STATES_AND_TERRITORIES,
map(lambda x: x.fips),
filter(lambda x: x is not None),
map(lambda x: int(x)), list)
SITECODE_TRANSLATORS = {
'fips':
lambda x: (us.states.lookup('%.2d' % x).abbr
if int(x) in US_STATES_FIPS_INTS else 'NA'),
'codes':
identity
}
SVYDESIGN_COLS = ['sitecode', 'strata', 'psu', 'weight']
def convert_cat_codes(s, fmt):
import click
import survey_stats
from survey_stats.dbi import DatabaseConfig, DatabaseType
import survey_stats.const as c
from survey_stats import log
logger = log.getLogger('cli')
def resolve_db_args(db_host, db_port, db_type,
db_user, db_password, db_name, db_config):
db_type = DatabaseType(db_type)
dbc = DatabaseConfig(host=db_host, port=db_port,
type=DatabaseType(db_type), user=db_user,
password=db_password, name=db_name)
if db_config:
dbc = DatabaseConfig.from_yaml(db_config)
return dbc
database_params = [
click.option('-c', '--cache-dir', type=c.CLICK_DIR_PATH,
default=c.DEFAULT_CACHE_DIR,
help='directory with data files default:cache'),
click.option('-C', '--db-config', type=c.CLICK_FILE_PATH,
help='database config yaml, takes priority'),
click.option('-H', '--db-host', type=click.STRING,
envvar='SVY_DBHOST', default='localhost',
help='hostname/ip for database'),
click.option('-P', '--db-port', type=c.CLICK_TCP_PORT,
envvar='SVY_DBPORT', default=50000,
def __init__(self):
self.logger = log.getLogger('statsworker.' + __name__)
from rpy2 import robjects as ro
from survey_stats.helpr import svyciprop_xlogit, svybyci_xlogit, factor_summary
from survey_stats.helpr import filter_survey_var, rm_nan_survey_var, svyby_nodrop
from survey_stats.helpr import fix_lonely_psus
from survey_stats import pdutil as u
from survey_stats.const import DECIMALS
from survey_stats import log
import gc
rbase = importr('base')
rstats = importr('stats')
rsvy = importr('survey')
rfeather = importr('feather', on_conflict='warn')
logger = log.getLogger()
def dim_design(d):
return pandas2ri.ri2py(rbase.dim(d[d.names.index('variables')]))
def subset_survey(des, filt, qn=None):
# filt is a dict with vars as keys and list of acceptable values as levels
# example from R:
# subset(dclus1, sch.wide=="Yes" & comp.imp=="Yes"
if not len(filt.keys()) > 0:
# empty filter, return original design object
return des
filtered = rbase.Reduce(
"&",
[filter_survey_var(des, k, v) for k, v in filt.items()] +
