Description

In data analysis you could examine the following topics:

Analyze Data Structures

Are data structures suited to represent the problem domain?

At first, make the structure of the existing data explicit, e.g. by creating a rough sketch of a data model as either informal diagrams, entity-relationship or class diagrams. Focus should be on overview: Where and how are which kinds of data stored in which format. What are the relationships between the data elements?

Second, create an explicit model of the required domain data structures.

Some typical questions might help in finding problems:

  • structural differences between those two models?
  • differences in data types?
  • differences in plausibility or validity checking?

Analyze Data Access

Get an overview of data access paths: How is data read or written? Do the queries match their requirements, or are complex mappings or unsuitable indirections involved?

  • What queries or executed how often?
  • How large are the results in number or volume?
  • Do relationships between query results have to be computed or do appropriate indices exist?

Analyze Data Size

  • Are some parts of the data especially large?
  • Is the relation between record-size (how large is a single record?) and record-volume (how many records exists?) plausible?
  • Do critical queries involve especially large parts of data?

Analyze Data Validation

  • How is data validated? (upon write, upon read, on client, on server, redundantly, uniformly)
  • Is validation consistent with current business rules?
  • Is validation overly complex?
  • Is validation implemented with appropriate technical means?

Analyze Data Actuality and Correctness

Especially in data concerning dynamic entities like people, organizations, markets, commodities etc., facts are very likely to change over time. Such data (stored facts) might become invalid sooner or later. Other types of information (like tax records, invoices or bookings on bank accounts) are created once and remain valid forever).

  • Tip: Peoples’ address typically changes something between 2-10 times during their lives.
  • Empirical studies show that between 5 and 10% of business or job email addresses become invalid every year.
  • Which parts of the data are subject to (what kind of) changes?
  • Are parts of the data known to be invalid or contain invalid portions?
  • Does the System handle potentially wrong or invalid data appropriately?
  • Are there (organizational or technical) processes in place that deal with data inconsistencies or faults?

Analyze Data Access Protection

  • Is there an overview what kinds of data need which level of (access) protection?
  • Is there a security concept in place covering protection against unauthorized access?
    • How are users/roles/organizations allowed to access data managed?
    • Is there a process in place to revoke access for outdated users/roles/organizations?
  • Is there a plan what shall happen in case of security breaches or data theft?
  • How is data theft recognized?

Analyze Backup

  • Is there a universal backup strategy in place, covering all areas of data storage?
  • Does the backup strategy match the business criticality of the data?
  • To what extend has the backup been verified?
  • Does a fallback scenario exist in case of (partial or complete) data loss?