Business Intelligence and Analytics (BI&A) is the process of extracting and predicting business-critical insights from raw data. Traditional BI focused on data collection, extraction, and organization to enable efficient query processing to derive insights from historical data. With sources of data sources growing steadily, traditional BI&A are evolving to provide intelligence at different scales and perspectives: operational BI, situational BI, self-service BI. In these slides we digested an article that talking about next generation of BI.
Table of Contents
1. Introduction
2. Traditional BI
3. Modern Features of Next BI Generation
3.1. Operational (Real-Time) BI
3.2. Situational Awareness BI
3.3. Self-Service BI
4. Data Architecture
4.1. Traditional BI Systems
4.2. Nowadays BI Systems
4.3. Next Generation BI Systems
5. Next Generation
5.1. Integrative Analysis
5.1.1. Multiple Kernel Learning
5.1.2. The Bayesian Network
6. Conclusion
Research Objectives and Key Topics
This work explores the evolution of Business Intelligence (BI) from traditional, rigid frameworks to modern, next-generation architectures that leverage machine learning to process heterogeneous and high-velocity data. The primary objective is to address the challenges of traditional storage and analytics by proposing integrative analysis techniques for more proactive business decision-making.
- Transition from traditional BI to real-time and self-service analytics.
- Analysis of modern data architecture challenges and solutions (HyPer, MobiDB).
- Integration of machine learning techniques for data fusion.
- Exploration of Multiple Kernel Learning and Bayesian Networks in BI.
Excerpt from the Book
Challenge for Existing Storage and Analytic Solutions
In today’s enterprise working world, the data is: heterogeneously generated, at a much faster rate, characterized by huge data sets, and varied data types where the nature of data being structured, semi-structured and unstructured as flat files, cubes, videos, images, audio, weblogs, text, and e-mail etc. This poses a challenge to the existing traditional storage and analytic solutions.
Chapter Summary
Introduction: Defines Business Intelligence as the process of extracting critical insights from raw data and highlights the shift in data sources from simple internal records to diverse modern streams.
Traditional BI: Discusses the classic 3-layer architecture and identifies core limitations such as slow performance, rigidity, and time-consuming processes.
Modern Features of Next BI Generation: Outlines the necessity for real-time operational data processing, situational awareness through external data, and user-friendly self-service platforms.
Data Architecture: Analyzes the evolution of technical systems from standard ETL and OLAP/OLTP setups to modern mechanisms like HyPer and MobiDB that handle mixed workloads.
Next Generation: Introduces advanced machine learning integration, specifically focusing on Multiple Kernel Learning and Bayesian Networks to achieve proactive analytical responses.
Conclusion: Summarizes the progression of BI architectures and the imperative for adopting intelligent, machine-learning-driven approaches to handle modern data complexities.
Keywords
Business Intelligence, Data Analytics, Real-Time BI, Self-Service BI, Data Architecture, OLAP, OLTP, Machine Learning, Integrative Analysis, Multiple Kernel Learning, Bayesian Network, Data Integration, Unstructured Data, Predictive Analytics, Enterprise Data.
Frequently Asked Questions
What is the core focus of this work?
This work focuses on the evolution of Business Intelligence (BI) technologies, moving from legacy, slow systems to agile, next-generation frameworks capable of handling modern, heterogeneous data.
What are the central themes of the research?
The central themes include the transition from traditional BI to real-time operations, the architectural challenges of modern data systems, and the application of machine learning for smarter insights.
What is the primary objective of the document?
The primary objective is to illustrate how integrating machine learning techniques, such as Multiple Kernel Learning and Bayesian Networks, can solve the limitations of current BI storage and analysis methods.
Which scientific methods are primarily discussed?
The research discusses system design methods for data handling (like lock-less snapshots) and statistical/algorithmic methods for integrative analysis, specifically MKL and Bayesian Networks.
What subjects are covered in the main body?
The main body covers the limitations of traditional BI, the requirement for real-time and situational awareness, modern system architectures (HyPer, MobiDB), and future-oriented integrative analysis models.
How would you characterize this work using keywords?
The work is characterized by terms like Business Intelligence, Real-Time BI, Data Architecture, Machine Learning, and Integrative Analysis.
What specific challenges do traditional BI systems face?
Traditional systems suffer from being too slow, rigid, and time-consuming, particularly when managing modern data sets that are diverse in type and generated at high velocity.
How does the "HyPer" system improve BI architecture?
HyPer improves performance by handling mixed OLTP and OLAP workloads simultaneously using low-overhead differential snapshots and a lock-less approach for transaction processing.
What role does the Bayesian Network play in the proposed next-gen BI?
The Bayesian Network is used to map direct qualitative dependence relationships between data variables, providing quantitative insights into the strength of those dependencies.
- Quote paper
- Mohammad Javad Khademian (Author), 2019, A Presentation About Next Generation Business Intelligence and Analytics, Munich, GRIN Verlag, https://www.grin.com/document/504492
