Kafka tutorial - Kafka + H2O

See Realtime Machine Learning predictions with Kafka and H2O.ai.

Data

File: data/housedata.csv

Data types:

• date: time
• price: numeric
• bedrooms: numeric
• bathrooms: numeric
• sqft_living: numeric
• sqft_lot: numeric
• floors: numeric
• waterfront: enum
• view: enum
• condition: enum
• sqft_above: numeric
• sqft_basement: numeric
• yr_built: numeric
• yr_renovated: numeric
• street: string
• city: enum
• statezip: enum
• country: enum

Splits:

• 70% for training
• 20% for validation
• 10% for test

Model

• Type: Gradient Boosting Machine
• Training frame: 70%
• Validation frame: 20%
• Response column: price
• Ignored columns: date
• ntrees: 120

Kafka

$ kafka-topics --zookeeper localhost:2181 --create --topic housing --replication-factor 1 --partitions 4
$ kafka-topics --zookeeper localhost:2181 --create --topic predictions --replication-factor 1 --partitions 4
$ kafka-topics --zookeeper localhost:2181 --create --topic zipcodes --replication-factor 1 --partitions 4