< Back Home

Mock data of Medicines

Loading fake information about medicines comsumption. Based on Mock Data dataset of fake people, this dataset records information about generic drugs and patient number which links with people information. Besides, each record store id_company that is its simulated manufacturer. This id_company reference Mock Companies datafile .

This Script loads generated mock data of generic medicines: list of active substances, manufacturer company, and fake references to fake patients who would take the drug, into Cassandra table “mock-drugs” mock_drugs_imp.py within “examples_bis” keyspace, in the table

Dependencies

Some python packages are needed:

from cassandra.cluster import Cluster
from cassandra.policies import TokenAwarePolicy, RoundRobinPolicy
from cassandra.cqlengine.models import Model
from cassandra.cqlengine.management import sync_table
from cassandra.cqlengine import connection
from cassandra.cqlengine.columns import *

More information about python driver for Cassandra

Dataset structure

COLUMNS = ['id_drug','drug_name','id_company','id_patient']

Defined constant COLUMNS synthetize the structure of the CSV file mock_companies.csv

This file was generated with online freemium tool Mockaroo which is able to generate ramdom values into CSV format with several avalaible types.

I’ve choosed a set of frequently used types:

• id_drug autoincremental id (unique)
• drug_name name of active substances or medicine composition
• id_company pharmaceutical company which manufacture the drug
• id_patient patient who is being treated with this drug

More details about data template on Mockaroo

this is a denormalized dataset; many to many relationship in one key-value table

Loading data from file system to memory

I use python pandas library to load CSV file into memory and to be able to work with.

df = read_csv(os.path.abspath(FICHERO_DATOS),header=0,names=COLUMNS,quotechar='"',decimal=',',encoding=ENCODING)

read_csv is very flexible method within the powerful pandas library which empower you to make multiple things. Load data into Dataframe structure, that is similar with Spark Dataframe; an important difference is that the last one is a distributed dataset and in pandas is locally stored in memory, but is relatively easy to convert between each other, so you could work with both at your convenience.

Storing into Cassandra

Using this class you can do object mapping with records into Cassandra table

## Object Mapper
class MockDrugs(Model):
  __keyspace__ = KEYSPACE
  id_drug = Integer(primary_key=True)
  drug_name = Text()
  id_company = Integer()
  id_patient = Integer()

Simply with sync_table(MockDrugs) you can manage record persistence. Model descendant classes inherit a method to create records which will be posted into Cassandra table.

for ind, row in tqdm(df.iterrows(), total=df.shape[0]):
  MockDrugs.create(
    id_drug = ind,
    drug_name = row['drug_name'],
    id_company = row['id_company'],
    id_patient = row['id_patient'],
  )