Mining heterogeneous information networks
0 likes142 views
The document discusses the importance and applications of heterogeneous information networks (HINs) in various fields, such as customer order mining and healthcare service consumption. It highlights the growing adoption of graph databases among enterprises and the potential for graph analysis to differentiate organizations in data-driven decision-making. Case studies presented demonstrate the value of mining HINs for improving customer insights and predicting behaviors.
Mining heterogeneous information networks
- 1. Mining Heterogeneous Information Networks Ashwin Pingali CTO – Apps Consultants & Dzee Solutions ashwin@appsconsultants.com
- 2. Agenda Why Graphs Graphs / Networks - History Homogeneous Networks Heterogeneous Networks HINs – examples Mining HINs - Examples Other Case Studies – Forecasting / Recommendations
- 3. “Graph analysis is the true killer app for Big Data.” “Forrester estimates that over 25% of enterprises will be using graph databases by 2017.” “Graph analysis is possibly the single most effective competitive differentiator for organizations pursuing data- driven operations and decisions after the design of data capture.” “By the end of 2018, 70% of leading organizations will have one or more pilot or proof-of-concept efforts underway utilizing graph databases.” Why Graphs
- 4. Graphs / Networks - History Is it possible to walk with a route that crosses each bridge exactly once, and return to the starting point. In 1736, Leonhard Euler proved that it was not possible. A B C D A B C D1 2 3 4 5 6 7 1 4 2 3 5 6 7
- 5. Homogeneous Networks C12 C11 C13 C14 C14 C1 C9 C7 C4 C5 C3 C10 C8 C6 Characteristics • Single type of node • Single link type Examples • Network of computers • Network of people
- 6. Heterogeneous Networks Characteristics • Different types of nodes • Different types of relationships Examples • Influencing the decision maker Decision maker Sales rep Employee Consultant Employee Consultant Sales rep Sales rep Employee Best tool to represent heterogeneous networks
- 7. Heterogeneous Information Networks (HINs) Example 1: Customer Ordering Information Network
- 8. Heterogeneous Information Networks (HINs) Example 1: Customer Ordering Information Network
- 9. Heterogeneous Information Networks (HINs) Example 1: Customer Ordering Information Network
- 10. Heterogeneous Information Networks (HINs) Example 1: Customer Ordering Information Network
- 11. Mining HINs – Customer Order Mining • Searching for Similar customers based on past order behavior (frequency, items ordered) • Classifying customers (clustering) and then ranking customers within each classification • Predicting Relationships – For example: recommending products that they might be interested in. ROI Case Studies • Identify least profitable and most profitable transactions • Predict problematic transactions before they are fulfilled • Increase the revenue from customer base by targeted upsell.
- 12. Heterogeneous Information Networks (HINs) Example 2: HealthCare Service Consumption Network
- 13. Heterogeneous Information Networks (HINs) Example 2: HealthCare Service Consumption Network
- 14. Mining HINs – HealthCare Service Consumption • Identifying similar customers based on their actual health care service usage • Predicting health care service consumption over time • Modeling Risk for different health care groups
- 15. Other Case Studies • Hierarchical forecasting taking underlying components into account. • Behavioral patterns of customers as networks evolving over time • Using networks to calculate cycle times