3. DW Initiatives and Models for Financial Analysis

A DW is set of historical data, sometimes obtained from different sources, and its main purpose is to support management decisions. Its implementation is a complex process, namely when one's problematic situation is different from other known implementations. Thus, it is important to take into consideration the information needs of decision makers (requirement-driven approaches), but also what data are available as a source of information to populate the DW (data-driven approaches). These two perspectives, with the necessary adaptations lead us to a hybrid approach. This implies the needs of the management to rely on the capability of exploring available data, considering high volumes of data, such as in a big data scenarios or just looking to a way of organizing the information in a credible an consistent form.

Literature also refers several proposals on multidimensional modelling and data warehouse design. However, it seems, there is no consensus on modelling and design method, but a set of complementary methods. The Conceptual Data Model for Data Warehouse of Kamble, points to a uniform way of modelling multidimensional concepts, data warehouse design and aggregations. The influence of this model on our project is at a level of research, which individually, focus on a small part of the problematic situation. In fact, each model referred by Kamble focus on a particular approach about managing data.

In the research on the business domain of the project we stated that finding studies of DW to solve the needs of bank entities seems to be common. However, studies of DW to link the bank accounts to the financial area of a financial company seem to be a new problematic situation. To turn around this adversity, we conduct our study by researching conceptual models that could inspire the design and the implementation of DWs in a broad sense.

The schema of a DW lies on two kinds of elements: (i) facts and (ii) dimensions. Facts are related with the metrics of situations or events, and dimensions are used to analyse the results of the metrics, through the application of a set of arithmetic operations (counting, summation, average, ...).

Establishing an analogy with our case, the movements of bank accounts are the facts and, each fact is characterized by a monetary value and by the value of the situation (debt or credit). The monetary values can be aggregated by debt or credit. However, other dimensions can be obtained: balance by year, balance by month, debts higher than a certain value, credits from a specific entity, among other dimensions. So, the value of the movement is the connection between the types of the movement with the restriction that filters a set of movements. In the simplest way, this type of connection concerns the organization of facts with regard to dimensions. The model of Schneider, has brought the unification of the notion of fact and the notion of dimension, which become a value to future implementations of DWs. Another valuable characteristic is the similarity among model representation and semantic web. Like the "Conceptual Data Model" of Kamble, it allows to solve adversities based on the reasoning of the semantic web. Both are related with mathematical theoretic semantics grounded on standard ER semantics.