Published: 08 January 2019
Summary
Artificial intelligence/machine learning and the domain of the multiengine architecture of the logical data warehouse are complementary. Technical professionals working in data and analytics can increase business benefit by architecting their systems to take advantage of this natural synergy.
Included in Full Research
- Method 1 — Interface the AI/ML Service or Product to the LDW
- Method 2 — Invoking Machine Learning Within the DBMS or DMSA
- Invocation Example: Forecasting Using Linear Regression
- Integrating AI/ML Algorithms at the Operating System Level
- Method 3 — Using AI/ML to Help Manage the LDW Workload
- Using AI/ML to Boost Concurrency for Short Queries
- Using AI/ML to Automatically Tune an LDW Subsystem
- Method 4 — Profiling and Cleansing the LDW Input Data
- Automated Data Detection
- Data Cleansing Within the DBMS
- Method 5 — Deployment of Models and Results
- Deployment Through a Model Description Language
- Simple Deployment Using an ODS
- Robust Deployment Is Essential to Retraining and Recalibration of Models
- Deployment Using Multiple Languages, Tools and Analytic Engines
- Strengths
- Weaknesses
- Classification Using Support Vector Machines
- Finding Relationships Using Association Analysis
- Combining Free-Form Text Input and Structured Data
- Invoking Algorithms in the Data Lake With Spark ML and MLlib
- Expanded Detail on the Flight Data Cleanup Example