BI systems and data analysis
Emisfera develops business intelligence system and data analysis by adopting a working method that starts from from heterogeneous data available in the various applications that make up the company's IT system to get to the data warehouse, reports and information dashboards.
ERP systems currently in use in many organizations, often supplemented by other specialized applications, supports the so-called operational activity, or dealing to store, manage and process all the information to carry out daily activities. The operational management systems are primarily concerned to enter and update data. They are based on the database type OLTP (On-Line Transaction Process), or database in which the main task is to insert, delete and modify the data.
In any organization, such systems should be integrated with applications dedicated to decision support, related to the management, production and storage of information needed to support executives in their strategic choices. Such business intelligence (BI) systems are based on data analysis techniques. The BI systems are based on OLAP database (OnLine Analytical Processing System) into which a large amount of data, which must be mainly read and grouped.
A BI system must have the following characteristics:
- present data in a format that is easy to read and understand, and that makes extensive use of graphics;
- possibility to treat large volumes of data with almost instantaneous response times;
- ability to integrate the data from different sources, both internal and external to the company;
- keep history of changes undergone by the data, to allow historical analysis.
A business intelligence project
The main phases of a BI project are:
- Data collection. They can come from ERP systems, relational databases, applications or other files. They can reside on different platforms and can contain structured information such as tables and spreadsheets, or unstructured information such as text files, images and other multimedia information.
- Extraction, transformaton and loading (ETL). Extracting data from the sources, their transformation in a common format, and subsequent loading on the target OLAP database. The transformation serves to cleanse and integrate data and prepare them for their new location..
- The data warehouse construction (DWH). It is a large database in which the data is organized in such a way as to facilitate the analysis. The DWH maintains historical information and allows one to analyze them according to different sizes.
- Analysis: building reports and dashboards
A data warehouse construction
The DWH is built using a model (Dimensional Fact Model - DFM) that allows you to correlate the main numerical data you want to analyze, that is, the facts (eg. Turnover, margin ...) with various dimensions that this data can be analyzed (eg. time, geographic, by product ...).
The DWH designed so it can then be implemented using a relational database (one speaks in this case of R-OLAP), or specialist that is multi-dimensional (M-OLAP). In the case of R-OLAP database, the DFM model is mapped to a relational structure using a technique called star schema in which facts and dimensions are mapped to corresponding relational tables.
Emisfera is able to use both techniques, with different tools.
The process of extraction, transformation and loading (ETL)
The ETL process of extracting data from various sources and the data warehouse loading is often the most complex of the entire project. At this stage we should be solved conceptual problems of integration of data from different sources, including addressing issues relating to the cleaning of the data (data cleaning).
The last component of a BI project consists of a set of applications that allow you to analyze, use and see the information in the DWH, representing tabular and graphic data extracted. Reports are interactive in the sense that, once designed, can be calculated, if desired, according to new data in the DWH.
A particular type of report is made up of the dashboard, used to represent the set of key performance indicators (KPI) business, which provides the key performance indicators (customer knowledge, internal production process, financial aspects, etc.. ). A key requirement is the multidimensional view of data with the ability to quickly change the outlook of analysis and detail levels, taking advantage of the presence of hierarchies.
The data analysis allows to analyze information from the corporate information system. Data mining instead consists in a set of techniques and algorithms for the extraction of non-obvious information, previously unknown and potentially useful, contained within the DWH. Data mining allows you to make predictive analysis, fundamental in areas such as sales forecasts, customer analysis and market analysis.
Data mining is based on techniques such as identifying association rules, classification, clustering, and similarity search, which allow detection of hidden similarities into homogeneous data groups (eg sales data) to provide estimates in the presence of similar data (eg. sales forecasts on specific user targets).
Business intelligence in small and medium-sized enterprises
The term business intelligence often evokes complex and unsustainable investment projects for small and medium-sized enterprises. It is no longer so. The reduction in prices of software used and the consolidation of the techniques now make BI accessible to businesses of any size, and can pay for themselves in a short time mainly due to the higher level of knowledge that these projects can afford to get on the various aspects of their business.
Business intelligence projects implemented
- Trend reports for the product WhiteNet, used by the main employment agencies in Italy
- Web-BI project: collection and analysis of data from the web related to regenerative medicine
- Ergos research project: BI applied to the management of Asset
- Decision-making system for Infor consortium and associates
- Data analysis system to Emisfera soc.coop.
- Data analysis system to Digima srl
- Provision of courses BI in the enterprise