Tech Tips: Redis Search - Card Transactions Example

Summary

I'll be demonstrating Redis Search capabilities in a credit card transaction domain. All the data will be synthetically generated from the Faker module. Data will be stored as Hash sets in Redis. Subsequently, Redis Search will be leveraged to generate analytics on the data.

Architecture

Code Snippets

Data Generation

	merchants_provider = DynamicProvider(
	provider_name='merchants',
	elements=['Walmart', 'Nordstrom', 'Amazon', 'Exxon', 'Kroger', 'Safeway', 'United Airlines', 'Office Depot', 'Ford', 'Taco Bell']
	)
	categories_provider = DynamicProvider(
	provider_name='categories',
	elements= ['AUTO', 'FOOD', 'GASS', 'GIFT', 'TRAV', 'GROC', 'HOME', 'PERS', 'HEAL', 'MISC']
	)

	def generate_data(client, count):
	Faker.seed(0)
	random.seed(0)
	fake = Faker()
	fake.add_provider(merchants_provider)
	fake.add_provider(categories_provider)
	for i in range(count):
	tdate = fake.date_time_between(start_date='-3y', end_date='now')
	txn_record = {
	'acct_id': int(fake.ean(length=13)),
	'txn_id': int(fake.ean(length=13)),
	'txn_date': re.escape(tdate.isoformat()),
	'txn_timestamp': time.mktime(tdate.timetuple()),
	'card_last_4': fake.credit_card_number()[-4:],
	'txn_amt': round(random.uniform(1, 1000), 2),
	'txn_currency': 'USD',
	'expense_category': fake.categories(),
	'merchant_name': fake.merchants(),
	'merchant_address': re.escape(fake.address())
	}
	client.hset(f'{PREFIX}{txn_record["txn_id"]}', mapping=txn_record)

view raw cctxn_datagen.py hosted with ❤ by GitHub

Index Creation

	try:
	client.ft(IDX_NAME).dropindex()
	except:
	pass
	idx_def = IndexDefinition(index_type=IndexType.HASH, prefix=[PREFIX])
	schema = [
	TagField('txn_id', sortable=True),
	TextField('txn_date'),
	NumericField('txn_timestamp', sortable=True),
	NumericField('txn_amt'),
	TagField('txn_currency'),
	TagField('expense_category'),
	TextField('merchant_name'),
	TextField('merchant_address')
	]
	client.ft(IDX_NAME).create_index(schema, definition=idx_def)

view raw cctxn_index.py hosted with ❤ by GitHub

Sample Query

The query below aggregates total spend by category for those transactions with a dollar value >$500 in Dec 2021.

	request = AggregateRequest('(@txn_date:2021\-12* @txn_currency:{USD} @txn_amt:[(500, inf])')\
	.group_by('@expense_category', reducers.sum('@txn_amt').alias('total_spend'))\
	.sort_by(Desc('@total_spend'))
	result = client.ft(IDX_NAME).aggregate(request)

view raw cctxn_search.py hosted with ❤ by GitHub

Source

https://github.com/Redislabs-Solution-Architects/cctxn-examples

Sunday, April 23, 2023

Redis Search - Card Transactions Example