def __init__(self,
region_properties,
regions_to_compare=None,
data_dir=None,
no_download=False):
"""
:param region_properties: A dict containing the properties of regions to use when doing the similarity
computation. This is by default defined in region_properties.py, but can be adjusted.
:param regions_to_compare: Default is None. This is a list of region IDs against which we want to do the
similarity computation. If none specified, do this for all regions where we have _any_ data.
"""
self.client = BatchClient(API_HOST, ACCESS_TOKEN)
self._logger = get_default_logger()
if not regions_to_compare:
regions_to_compare = self._regions_avail_for_selection(
region_properties)
self._logger.info("SimilarRegionState: loading...")
self.state = SimilarRegionState(region_properties,
regions_to_compare,
self.client,
data_dir=data_dir,
no_download=no_download)
self._logger.info("SimilarRegionState: done.")
self._logger.info("BallTree: computing...")
# Featues are weighted at this point, so this is equivalent to using
# sklearn.neighbors.DistanceMetric.get_metric('wminkowski', **{'p':2,'w':self.state.weight_vector})
# but with unweighted features
self.ball = BallTree(self.state.data_standardized, leaf_size=2)
self._logger.info("BallTree: done.")
return
def __init__(self,
region_properties,
regions_to_compare=None,
data_dir=None):
"""
:param region_properties: A dict containing the properties of regions to use when doing the similarity
computation. This is by default defined in region_properties.py, but can be adjusted.
:param regions_to_compare: Default is None. This is a list of region IDs against which we want to do the
similarity computation. If none specified, do this for all regions where we have _any_ data.
"""
self.client = BatchClient(API_HOST, ACCESS_TOKEN)
self._logger = get_default_logger()
if not regions_to_compare:
regions_to_compare = self._regions_avail_for_selection(
region_properties)
self._logger.info("SimilarRegionState: loading...")
self.state = SimilarRegionState(region_properties,
regions_to_compare,
self.client,
data_dir=data_dir)
self._logger.info("SimilarRegionState: done.")
self._logger.info("BallTree: computing...")
self.ball = BallTree(self.state.data_standardized, leaf_size=2)
self._logger.info("BallTree: done.")
return
def __init__(self, api_host, access_token):
super(BatchClient, self).__init__(api_host, access_token)
self._logger = lib.get_default_logger()
self._http_client = AsyncHTTPClient()
def __init__(self, api_host, access_token):
super(GroClient, self).__init__(api_host, access_token)
self._logger = lib.get_default_logger()
self._data_series_list = set() # all that have been added
self._data_series_queue = [] # added but not loaded in data frame
self._data_frame = pandas.DataFrame()
