Data Warehousing
& Data Mining
& Data Mining
Wolf-Tilo Balke Silviu Homoceanu
Institut für Informationssysteme
Technische Universität Braunschweig http://www.ifis.cs.tu-bs.de
13. Decision Support Systems (DSS)
13.1 Marketing Models
13.2 Supply Chain Management
13. Decision Support Systems
• Decision support systems (DSS)
– Are interactive, flexible, and adaptable content based information systems
– Developed for supporting the solution of a non-
structured management problem for improved
13.0 DSS - Introduction
structured management problem for improved decision-making
– It utilizes data, it provides easy user interface, and it allows for the decision maker’s own insights
– DSS evolve as they develop
– The support for the decision layer is provided by traditional approaches, data mining and data warehousing with OLAP
13.0 DSS - Introduction
• Traditional approaches
– Common mathematical modeling e.g., what-if-analysis – Non-rigorous modeling
• Data-driven
– Rule-based systems (RBS)
• Data Warehousing
13.0 DSS - Introduction
• Data Warehousing
– Online Analytical Processing (OLAP)
• Data-based decision support
– Modeling
• Conceptual modeling
• Logical modeling
• Physical modeling
– ETL-Processes
• Data Mining
– Association rule mining
– Sequence patterns and time series – Classification
13.0 DSS - Introduction
Classification – Clustering
• Decision-making is a process of making the choice including
– Assessing the problem
– Collecting and verifying information – Identifying alternatives
13.0 DSS - Introduction
– Identifying alternatives
– Anticipating consequences of decisions
– Making the choice using sound and logical judgment based on available information
– Informing others of decision and rationale – Evaluating decisions
• Decision problem
13.0 Decisions
options (alternatives)
goals
• FIND the option that best satisfies the goals
• RANK options according to the goals
• ANALYSE, JUSTIFY, EXPLAIN, …, the decision
• Types of decisions
– Easy (routine, everyday) vs. difficult (complex)
– One-time vs. recurring – One-stage vs. sequential
13.0 Decisions
– One-stage vs. sequential
– Single objective vs. multiple objectives – Individual vs. group
– Structured vs. unstructured – Tactical, operational, strategic
• DSS address complex decisions
• Characteristics of complex decisions
– Novelty
• There was no prior similar decision
– Unclearness
• Incomplete knowledge about the problem
13.0 Complex Decisions
Incomplete knowledge about the problem
– Uncertainty
• Outside events that cannot be controlled
– Multiple objectives (possibly conflicting)
• Maximize economic benefits vs. minimize environmental costs
– Group decision-making
Important consequences of the decision
• Decision-making (DM)
13.0 Decision-Making
Machine DM Human DM
Decision Systems
• Switching circuits
• Processors
• Computer programs
• Systems for routine DM
• Autonomous agents Decision Sciences
• Decision-making
13.0 Decision-Making
Decision Systems
Decision Sciences
Normative Descriptive Decision Theory
Utility Theory Game Theory
Theory of Choice
Cognitive Psychology Social and Behavioral
Sciences
…
Automated Control Fuzzy Logic
Expert Systems
…
• Decision support
– Methods and tools for supporting people involved in the decision-making process
– Central Disciplines
Operations research and management sciences
13.0 Decision Support
• Operations research and management sciences
• Decision analysis
• Decision support systems
• DSS capabilities
– Support for problem-solving phases
• Intelligence, design, choice, implementation, monitoring
– Support for different decision frequencies, e.g.:
13.0 DSS - Introduction
• Ad hoc DSS: decisions that come up once in every 5 years (e.g., where should a company open a new distribution
center?)
• Institutional DSS: decisions that repeat (e.g., what should the company invest in?)
– Support for different problem structures
• Highly structured problems: known facts and relationships
• Semi-structured problems: facts unknown or ambiguous, relations vague
– E.g., which person to hire for a position?
13.0 DSS Capabilities
– E.g., which person to hire for a position?
– Support for various decision-making levels
• Operational level
– Daily decisions
• Tactical level
– Planning and control
• Strategic level
• DSS architecture
– Information resources – The analytical engine – The user interface
13.0 DSS - Introduction
Model management
External models
DW
Graphical User Interface Knowledge-based
subsystem
• The database management subsystem
– Captures/extracts data for inclusion in a DSS database – Updates (adds, deletes, edits,
changes) data records and files
13.0 DSS Architecture
records and files – Interrelates data
from different sources
– Retrieves data from the database for
queries and reports
– Provides comprehensive data security (protection from unauthorized access, recovery capabilities, etc.) – Handles personal and unofficial data so that users
can experiment with alternative solutions based on their own judgment
13.0 DBM Subsystem
their own judgment
– Tracks data use within the DSS
– Manages data through a data dictionary
• The model management subsystem (MMS)
– Strategic models: non routine mergers, impact analysis, capital budgeting
– Tactical Models:
allocation & Control labor requirements, sales promotion
13.0 DSS Architecture
sales promotion planning
– Operational Models:
routine-day-to-day
production scheduling, inventory control,
quality control
– Analytical Models:
SPSS, data mining
• Major functions of the MMS
– Creates models easily from scratch or from existing models
– Allows users to manipulate models so that they can conduct experiments and sensitive analysis e.g., what-
13.0 MMS
conduct experiments and sensitive analysis e.g., what- if or goal seeking analysis
– Manages and maintains the model base e.g.,
• Store, access, run, update, link, catalog and query
• The knowledge based subsystem
– Component of more advanced DSS
– Provides expertise in solving complex
unstructured and semi-structured problems
Expertise is provided by an expert system or other
13.0 DSS Architecture
• Expertise is provided by an expert system or other intelligent system
– Leads to intelligent DSS
– Example of knowledge extraction subsystem is data mining
• The user interface
– Interactive, dialogue oriented, menu driven – Intuitive, graphical, symbolic
– Consistent syntax and semantics, layout and
13.0 DSS Architecture
Consistent syntax and semantics, layout and symbolism
– Intelligent, context aware – Customized
• For the non-technical user, the user interface is
the system
• Applications of DSS
– Marketing Models
– Supply Chain Management
13.0 DSS - Introduction
• Marketing decision processes are characterized by a high level of complexity
– Simultaneous presence of multiple objectives
– Countless alternative actions resulting from the combination of the major choice options
13.1 Marketing Models
Countless alternative actions
combination of the major choice options
• Massive sales transactions data are available
making DSS a important tool for reaching
marketing intelligence
• Marketing intelligence comprises 2 prominent topics
– Relational marketing (RM)
– Sales force management (SFM)
13.1 Marketing Models
• Relational marketing as DSS application
– Designed to create, maintain, and enhance strong
relationships with customers and other stakeholders – Application of predictive models to support
relational marketing strategies
13.1 Marketing Models
relational marketing strategies – E.g.:
• An insurance company wishes to select the most promising market segment to target for a new type of policy
• A mobile phone provider wishes to identify those customers with the highest probability of churning
• Why is RM important?
– It costs five times as much to attract a new
customer as it does to keep a current one satisfied – It is claimed that a 5% improvement in customer
retention can cause an increase in profitability of
13.1 Relational Marketing
retention can cause an increase in profitability of between 25-85% depending on the industry
– Likewise, it is easier to deliver additional products and services to an existing customer than to a first-time buyer
• RM strategies revolve around the following choices
13.1 Relational Marketing
Distribution Products
Relational marketing
Sales processes Distribution
channels Products
Services
Segments
• How do we implement RM?
– Using pattern recognition and machine learning models on a company’s DW it is possible to derive different segmentations of the customers which are then used to
13.1 Relational Marketing
then used to design and target marke- ting actions
• Cycle of RM analysis, phases:
1. Exploration of the data available for each customer
2. Identify market segments by using inductive learning models
3. Knowledge of customer profiles is then used to design marketing actions
13.1 Relational Marketing
marketing actions
4. The designed actions are
translated into promotional campaigns
which generate in turn new information for
subsequent analyses
Collect information on
customers
Identify segments and needs Perform optimized
and targeted actions
• General statistics show…
– The average business never hears from 96% of its unhappy customers
• 91% never come back
• Dissatisfied customers may tell 9-10 people about their
13.1 Customer Relations
• Dissatisfied customers may tell 9-10 people about their experience
– Every positive experience is told to 4-5 people
– For every complaint received the average business in fact has 26 customers with a similar concern
– Of the customers who register a complaint, as many as 70% will do business again with your organization, if the complaint is resolved effectively
• This figure goes up to 95% if the complaint has been resolved quickly
13.1 Customer Relations
resolved quickly
– 40% of complaints are the result from customer mistakes or incorrect expectations
– A complaint that is handled efficiently is actually better than no complaint at all
• Customers who complain and get satisfactory results are
• Important part of RM is customer relationship management (CRM)
• CRM
– The software tools which allow tracking and
analysis of each customer's purchases, preferences,
13.1 Customer Relations
analysis of each customer's purchases, preferences, activities, tastes, likes, dislikes, and complaints
– Enterprise vendors/products
• Oracle/Siebel, SAP, Salesforce.com, Amdocs, Microsoft Dynamics
– Open source tools
• Opentaps, Tunesta, Compiere, XRMS, SugarCRM
• E.g., XRMS
– Contact
information screen
13.1 Customer Relations
• Aspects of CRM systems
– Operational – Collaborative – Analytical
13.1 Customer Relations
• Operational CRM
– Provides support to "front office" business processes, including sales, marketing and service
– Each interaction with a customer is generally added to a customer's contact history, and staff can
13.1 CRM
to a customer's contact history, and staff can
retrieve information on customers from the database when necessary
– Main benefits is that customers can interact with different people in a company over time without having to describe the history of their interaction each time
• Collaborative CRM
– Covers aspects of a company's dealings with customers that are handled by various departments within a company
• E.g., sales, technical support and marketing
– Staff members from different departments can share
13.1 CRM
– Staff members from different departments can share information collected when interacting with customers
• E.g., feedback received by customer support agents can provide other staff members with information on the services and
features requested by customers
– Goal of collaborative CRM is to use information collected by all departments to improve the quality of services
• Analytical CRM
– Analyzes customer data for a variety of purposes:
• Design and execution of targeted marketing campaigns to optimize marketing effectiveness
• Design and execution of specific customer campaigns, including customer acquisition, cross-selling, up-selling, retention
13.1 CRM
customer acquisition, cross-selling, up-selling, retention
• Analysis of customer behavior to aid product and service decision making e.g., pricing, new product development
• Management decisions, e.g. financial forecasting and customer profitability analysis
• Prediction of the probability of customer defection (churn)
• Acquisition? Cross-selling? Up-selling? Retention?
• Lifetime of a customer
– Lost proposal
• Before becoming a customer, an individual may receive repeated proposals from the enterprise to win him/her as a customer
13.1 Relational Marketing
as a customer
– Acquisition
• The individual
becomes customer
– Cross/up-selling:
getting more business from current customers by selling them additional or complementary
services
– Retention:
13.1 Lifetime of a customer
– Retention:
the continuous attempt to satisfy and keep current customers actively involved in conducting business
• Highly satisfied customers are
– Less price sensitive
– More likely to talk favorably about you
– Churn (defection):
the percentage of customers who leave a business in one year
– Interruption:
customers leaving a business. Possible reasons are that
13.1 Lifetime of a customer
customers leaving a business. Possible reasons are that they:
• Die
• Move away
• Leave for competitive reasons
• Are dissatisfied
Quit because of an attitude of indifference
• Sales force management (SFM)
– Management of the whole set of people and roles that are involved with different tasks and
responsibilities in the sales process
• Why SFM?
13.1 Marketing Models
• Why SFM?
– It plays a critical role in:
• The profitability of an enterprise
• The implementation of the relational marketing strategy
• Designing the sales network and planning agents activities involve complex decision making tasks
– Remaining activities are operational and sales force automation (SFA) software can be used
• SFM decision-making process can be grouped in
13.1 Sales force management
• SFM decision-making process can be grouped in 3 components each interacting with each other
– Design – Planning
– Assessment
Sales force management
Planning Design
• Design
– During start-up phase or during restructuring – Includes 3 types of decisions
• Organizational structure
13.1 Sales force management
• Sizing
• Sales territories
– Organizational structure
• May take different forms corresponding to hierarchical agglomerations of agents by group, products, brand or geographical area
• In order to determine the organizational structure it is
13.1 Design
• In order to determine the organizational structure it is
necessary to analyze the complexity of customers products and sales activities
– Decide whether and to what extent the agents should be specialized
– Sizing
• Decide the number of agents that should operate in the selected structure
• Depends on several factors
– Number of customers, prospects, sales area coverage estimated
13.1 Design
– Number of customers, prospects, sales area coverage estimated time for each call, the agents traveling time, etc.
• Conflicting goals
– Reduction in costs due to decreasing sales force size is often followed by a reduction in sales and revenues
– Sales territories
• Deciding on assigning territories to agents
• Depends on factors such as
– The sales potential of the geographical areas
13.1 Design
– The sales potential of the geographical areas – The time required to travel from an area to
another
– The availability time of each agent
• Purpose of assignment is to determine a balanced situation between sales opportunities in each territory to avoid
disparities among agents
• Planning
– Decision-making process involving the assignment of sales resources structured and sized during design phase, to market entities
• E.g., sales resources
13.1 Sales force management
• E.g., sales resources
– Work time, budget
• E.g., market entities
– Products
– Market segments
– Distribution channels
• Assessment
– Measure the effectiveness and efficiency of the individuals in order to decide incentives and
remuneration schemes
• Define adequate evaluation criteria that take into
13.1 Sales force management
• Define adequate evaluation criteria that take into
account the personal contribution of each agent having removed effects due to area or product characteristics
• Sales Force Automation software
– Most CRM tools include SFA functionality – Enterprise vendors/products
• Oracle/Siebel, SAP, Salesforce.com, Microsoft Dynamics, Netsuite
13.1 Sales force management
Netsuite
– Open source tools
• XRMS, SugarCRM, Vtiger
13.1 Sales force management
• For producing industries, another field of business operation is of great importance:
– Supply chain management (SCM)
• A supply chain summarizes the logistic and
production processes of a single enterprise as
13.2 Supply Chain Management
production processes of a single enterprise as well as a network of companies
– Covers the flow of materials and products from the raw material down to the end product at the
customer
• Contains acquisition of raw materials,
• Within a single company, internal supply chain can be modeled and optimized
– Contain aspects of martial purchase, production and distribution
13.2 Supply Chain Management
Internal Supply Chain
Purchasing Production Distribution
Suppliers Customers
• However, global supply chains may form
complex networks of various material flows and costs
13.2 Supply Chain Management
European Plant
Recycling 1 European Assembly European Suppliers
Main Plant European Plant
US Assembly
Asian Market European Market Asian Assembly
European Assembly European Suppliers
US Suppliers
• Supply chain management is about managing
and optimizing those complex supply networks
– Eliminating excess inventory
– Improvise on-time delivery performance
13.2 Supply Chain Management
on-time delivery
– Maximize the value of procurement – Minimize transport costs
– Minimize storage costs – Etc.
• Steps of SCM
– Plan (strategic portion of SCM)
• Strategy for managing all the resources that go toward meeting customer demand
• Developing a set of metrics to monitor the performance of the supply chain so that it is efficient, costs less and delivers high quality
13.2 Supply Chain Management
– Source
• Choose suppliers to deliver the goods and services
• Develop a set of pricing, delivery and payment processes with suppliers
• Create metrics for monitoring and improving the relationships
• Put together processes for managing goods and services inventory, including receiving and verifying shipments, transferring them to the
– Make (manufacturing step)
• Schedule the activities necessary for production, testing, packaging and preparation for delivery
• Most metric-intensive portion of the supply - measure quality levels, production output and worker productivity
– Deliver (the logistics part)
13.2 Supply Chain Management
– Deliver (the logistics part)
• Coordinate the receipt of orders, develop a network of warehouses, pick carriers to get products to customers and set up an invoicing system to receive payments
– Return
• Receive and manage defective or excess products
• Recycle used products
13.2 Supply Chain Management
• For solving these tasks, SCM has to span across most other enterprise management areas
– Thus, software
solutions are usually very diverse and
Supply Chain Strategy
very diverse and customized
– Highly dependent on data from
all branches of business
Supply Chain Management
Supply Chain Planning
Supply Chain Enterprise Applications Product
Lifecycle Management
Logistics
• The traditional approach for optimizing supply chains was severely hampered by the unavailability of
necessary data
– Thus, usually only future demand was forecast as good as possible, using statistical trending and “best fit”
techniques
13.2 Supply Chain Management
techniques
• Only high level data necessary
– e.g. by weekly data by product category and customer group
• For dealing with unpredictability, security margins are added
• Based on the estimates, the supply chain could be optimized
– Capacity Planning
– Bill of Material problems – Network flow optimization – etc.
• However, due to improved data warehouse strategies, more dynamic and fine-grained optimizations are possible
– Forecasting at much finer-granularity
13.2 Supply Chain Management
• e.g. calculate the best inventory level per article for each store
• So called model stock
– Allows for new optimization techniques
• Simulation
– Include wider verity of metrics
• Stackability constraints
• Load and unloading rules
• Palletizing logic
• Warehouse efficiency
13.2 Supply Chain Management
• Warehouse efficiency
• “Shipping air” minimization
• Mondrian
– Open source OLAP engine provided by Pentaho – Based on ROLAP technology
– Is able to work with any major DBMS
13.3 The Mondrian System
Is able to work with any major DBMS
• Terradata, Oracle, IBM DB2, Sybase, Microsoft SQL Server, Microsoft Access, MySQL, Informix, PostgreSQL, etc.
• http://is59.idb.cs.tu-bs.de/mondrian/
The End
• I hope you enjoyed the lecture and learned at least some interesting stuff…
– Next semester’s master courses:
Multimedia Databases, XML Databases, GIS