The data used for the business analysis will be organized in special storages (data warehouse, DW). These data should reflect the current, real and broad picture of business. Information (including historical data) is going to the data warehouse from the different operating (transaction) systems and is structured in a special way for more effective analysis and request processing (in an appearance from normal databases where information is organized so that to optimize time of processing of current transactions).
Data warehouses contain huge information volumes, covering all available aspects of activity of the enterprise and all aspects of functioning of business allowing to consider in total. For the solution of narrower, specific objectives subsets of data – so-called data marts (data marts) can be isolated from the general storage.
There are several approaches to data integration.
For formation and maintenance of data warehouses so-called ETL means – extraction tools of these (extract), data translations (transform), i.e. their reduction to a necessary format, processings according to certain rules, combination with other data, etc. and also for data loading (load), data writing in storage or in other base are used.
In addition to ETL, BI systems include tools for work with SQL (structured query language) allowing users to address directly data. Recently tools for formation and request processing became more "friendly", focused on unprepared business users (but not on the qualified IT specialists).
For data integration from separate sources in the modern BI systems the interface, virtual layer of metadata is used that allows to save business users from need to deal with subtleties of storage and information processing and facilitating making changes. These means do not require any physical operations on movement and data processing that distinguishes them from ETL tools. Use of a similar layer of metadata, in principle, allows to refuse the organization of expensive data warehouses (however at the same time it is necessary to take questions of ensuring necessary performance into account). Such approach to integration of analytics from TEC (Technology Evaluation Centers) is defined as EII (enterprise information integration).
Besides, for data integration the corporate portals providing interrelation at the level of data and business processes can be created. Such portals implement only external interrelation, in other words – provide shared access to information. Such implementation at experts from TEC received the name EAI (enterprise application integration).
For comprehensive data analysis in modern BI the OLAP tools (online analytical processing) is used. They allow to consider different cuts of data, including the temporary, allowing to reveal different trends and dependences (on regions, products, clients, etc.).
For data view different graphic means – the reports, diagrams, charts configured by means of parameters are used.
The most developed BI solutions include blocks for an in-depth study of data (data mining). Sometimes this term is mistakenly used for designation of the tools allowing to provide in a new way (to display) information, however actually these tools are designed to help with detection of the hidden (unevident) patterns, models, drawing up forecasts. They are based on scanning and statistical processing of huge data arrays and finally are designed to facilitate acceptance of the correct and justified strategic decisions thanks to the analysis of different options of succession of events. As tools neural networks, decision trees are used.
Panels and cards
The commonly accepted means of data visualization in modern BI solutions are information (control, instrument) panels (dashboards) on which results are displayed in the form of the scales and indicators allowing to control the current values of the selected indicators, to compare them to critical (minimum \maximum allowed) values and thus to reveal potential threats for business.
Control panels are considered as one of the most convenient methods of information representation about "state of health" of business. They allow to find room on the screen for all major information on the current transactions revealed and potential problems.
Control panels, as well as score cards (scorecards), are based on the analysis of key performance indicators (KPIs). However, as a rule, control panels display a current status of the general indicators, and score cards are intended for comparison of the current indicators with planned, target, and display dynamics of change of these indicators in time. Score cards usually are more personalized, are configured depending on roles and tasks of the specific user (financial management, supply, sales, etc.). If necessary all these indicators can be detailed by means of additional reports, diagrams and charts.