config.yaml

# Attribute Selection algorithm pipeline 


# Attribute on the basis of which blocking needs to be performed (Very important with respect to performance)
# Needed while sample generation and generation of processed data.
# The number should be between

BLOCKING_ATTRIBUTE: 3

##
##
##

COMPARISON_ATTRIBUTE: [1,2,3,4,5,6,7,8,9,10,11,12,13,14]

##
## Sampling Rate (how much should be the size of sample with respect to the unique records in it)
##
SAMPLE_RATE: 0.02

##
## This is used to allow errors while identifying records. What is threshold tolerable while perfoming the linking.
##

THRESHOLD: 2



# Input data files. (Please place them in the data folder)


input_files:
  input_file_1: "../data/pse_dec.1.1"
  input_file_2: "../data/pse_dec.1.2"


#
# Names of the Attributes and the location in the dataset.
#

header: ['simulant_id', 'first_name', 'middle_initial', 'last_name', 'age',     
       'date_of_birth', 'street_number', 'street_name', 'unit_number', 'city',
       'state', 'zipcode', 'relation_to_reference_person', 'sex',
       'race_ethnicity']

#
# Unique identifier for the data
#

id_column: "simulant_id"

#
# Names of the files generated after sampling (NO NEED to change this)
#

sample_output:
  output_file_1: '../data/out.1.1'
  output_file_2: '../data/out.1.2'


#
# Name of the configuration file for RLA code (NO NEED to change this)
#

rla_xml_file: "../data/config_test.xml"

# OPTIONAL
# The extraneous data will get pruned. This parameter gives freedom to specofy amount of duplicate records (extraneous) can be there
# Suggested value between 0.2 to 0.3


FEATURE_PRUNE_KEEP: 0.2


##
## NO NEED to change this.
##

pruned_output: "../data/pruned_out.csv"


##
## NO NEED to change this.
##

rules_output: "../data/rules_out.csv"


##
## Stages of the PIPELINE
## Can enable disable stages. Each stage requires output from previous stage (Please take note of this).
##

run:
  sample_generation: False
  processed_data_generation: False
  feature_prune: False
  attribute_selection: True