Analytics infrastructure, platforms and methods
Post on 09-Aug-2015
81 Views
Preview:
Transcript
1
Analytics – Infrastructure, Platforms and Methods.
Feyzi Bagirov26 Jan 2015
2
Only about half of financial services organizations have a single system to comply with anti-money laundering directives (Accenture)
http://www.forbes.com/sites/jasonbloomberg/2014/07/29/three-way-big-data-banking-battle-brewing/
3
Data Mining ◦ Retail Use cases◦ Data Mining Process
Data Mining Methodologies
Data◦ Data Training◦ Types of Business Information Systems◦ Data Warehouses◦ Data Mining Tools◦ Data Visualization Tools◦ Big Data
Data Mining
A
B
CD
E
4
Machine Learning is a scientific discipline that explores the construction and study of algorithms that can learn from data (Ron Kovahi; Foster Provost (1998). “Glossary of terms”.
Data Mining is the process of achieving Machine Learning.
What is Data Mining?
A
B
CD
E
5
Response modeling for direct marketing Uplift modeling for direct marketing Customer retention with churn modeling Churn uplift modeling
Retail Use Cases
A
B
CD
E
6
Use Case 1 – Response Modeling For Direct Marketing
Lifeline Screening: Response up 38%, cost down 20%, 62K more customers annually
PREMIER Bankcard: Direct mail response up 3-5%
Sun Microsystems: Doubled the number of leads per phone call
A
B
CD
E
Based on the past experience, who will respond tomorrow?
7
Use Case 2- Uplift Modeling for Direct Marketing
A
B
CD
E
8
Use Case 2- Uplift Modeling for Direct Marketing
A
B
CD
E
Leading financial institution: incremental conversion up 0.02% to 0.43%; Revenue per contact up by over 20 times
9
Use Case 3 – Customer Retention With Churn Modeling
Reed Elsevier’s Caterer & Hotelkeeper: Reduced churn by 16%; Retention ROI up by 10%
PREMIER Bankcard: $8 million est. retained
Leading North American Telecom: Identified customers with a 600% increased risk of churn with social network analysis.
Optus (Australian telecom): Doubled churn model performance with social data
A
B
CD
E
10
Use Case 4 – Churn Uplift Modeling
A
B
CD
E
Telenor: Reduced churn 36%; Cost-of-contact down 40%; Campaign ROI up 11-fold
US Bank: Costs down 40%, lift up 2 times, and cross-sell ROI up 5 times
11
Data Mining ProcessCRISP-DM
(Cross Industry Standard Process for Data Mining)
SEMMA(Sample, Explore, Modify, Model, Assess)
A
B
CD
E
12
Business Task
Data Set
Data Preparation
Data cleaning
ModelingEvaluation
and validation
Use of DM results/deplo
yment
Results of action based
on DM results
Development
Data Mining Process
Strategic Objectives
Operational
Objectives
Marketing Objectives
Other Objectives
A
B
CD
E
13
Supervised (data training) and unsupervised methods
Age: 25-35Gender: MaleMarital Status: MarriedEducation: Graduate
Historically
Historically
Training Data
Unknown DataPrediction
Superv
ised
Unknown Data
Historically
Unsu
perv
ised
14
Transactional vs. Analysis-Based Systems
Transactional Information Systems
Analysis-Based Information Systems
15
Data Warehouses
Data Warehouse 4 main features:• Topical Orientation (customer, product, etc.)• Logical integration and homogenization (relational integration)• Presence of a reference period (vs operational)• Low volatility (should not change often)
3 components of Data Warehouses:• DBMS (Database Management System)• DB (Database)• DBCS (Database Communication System)
Snowflake Star
16
Data Marts
17
Data Mining Tools
18
Data Visualization Tools
19
What is Big Data?
1. Velocity2. Variety3. Volume
20
Q&A?
21
top related