Upload
dataversity
View
918
Download
1
Embed Size (px)
DESCRIPTION
Data Quality begins with the conceptual model. It's imperative that the modeler not only acknowledges data must be quality to be useful, but that they follow that paradigm all the way thru. It's about considering what you want the outcome of data to be, not only what you want the quality to be going in. Sue will share some ideas on how she has modeled in the past with an eye to quality, how she has got the business to provide the quality attributes and how she has managed to separate the mandatory from the nice to have. Attendees should come prepared with questions targeted at issues or concerns they are currently facing or have faced in the past.
Citation preview
Copyright 2014 by EPI-USE Data Services
October 2014
Data Quality for Data ModellersSue Geuens CDMP, MDQM
Copyright 2014 by EPI-USE Data Services
Data Quality Management is a critical support
process in organisational change management
Data Quality is synonymous with information
quality, since poor data quality results in
inaccurate information and poor business
performance
Data Quality is a LONG TERM
Program, not a SHORT TERM project
Copyright 2014 by EPI-USE Data Services
Data Quality is … and isn’t …
• Supposed to improve your
data
• Required to ensure that reports
have appropriate output
• Needs to enable your
executives to make the correct
decisions
• Must be assessed before any
migration/ integration project
• DOCUMENTED
• A once off instance of
cleansing a piece of data
• Supposed to fix the errors
created by incorrect data
modelling
• Going to improve without
concerted effort
• GUNG HO effort that dies
Copyright 2014 by EPI-USE Data Services
Interface Examples
Copyright 2014 by EPI-USE Data Services
Copyright 2014 by EPI-USE Data Services
Copyright 2014 by EPI-USE Data Services
Copyright 2014 by EPI-USE Data Services
What does Dilbert say?
Copyright 2014 by EPI-USE Data Services
Data Model Examples
Copyright 2014 by EPI-USE Data Services
Copyright 2014 by EPI-USE Data Services
Copyright 2014 by EPI-USE Data Services
Copyright 2014 by EPI-USE Data Services
Reasons for No Quality in Models• Cost
• Timelines
• Access to Data
• Culture
• Metadata
• Over Optimistic on current model
• Measures
• Business Process does not require Quality
• Data Flows
• Not in Your Scope
Copyright 2014 by EPI-USE Data Services
What is your Data Quality Maturity Rating?
Copyright 2014 by EPI-USE Data Services
Dimensions of Quality• Accuracy
Degree to which data correctly represents “real-life” entities
• Completeness Level of assigned data values that are required by business, system, application
• Consistency Applies to ensuring data sets across systems are consistent and/ or not in conflict
• Currency How “fresh” is the data compared to length of time last refreshed
• Precision Level of detail in the data element requiring specific accuracy
• Privacy Need for access control and usage monitoring
• Reasonableness Consider consistency expectations in systems and applications
• Referential Integrity Level to which data is related across database tables and columns
• Timeliness Availability of data for use and ease of accessibility
• Uniqueness The level to which the data entity is unique in the data set
• Validity Conformance to data element attributes, may be specific to database, system and/ or application
Permissable Purpose