Data Mining

What is Data Mining

Data Mining means extrapolating unknown information from within a large database. Such process can identify and provide the outcome of relations, schemes or anomalies which would not be visible through traditional extraction. A conventional extraction procedure would take much more time, without reaching the pre-set objective.

The traditional extraction approach, substantially featuring an initial question and information research mode throughout the data base, does not add any value as extraction algorithms are limited to highlight research outcomes. On the contrary, data mining is programmed to ‘capture’ through particular algorithms an information set which is useful to recover recurring partners or evident anomalies, but also some exploitable data at strategic and commercial level.

Data mining can include a definite number of steps:

• Selection of the data to be processed and the metrics to be returned;
• Analysis of the outcomes;
• Display of the disclosure or reporting.

Each step is of great value, because it contributes to achieve the ultimate result.

The quality of the data provided at the beginning of the extraction (data governance) is therefore of paramount importance for the positive completion of the process.

If those data turn out to be incorrect, data mining would not be able to extrapolate the required information and the data processed and highlighted in the reporting would not be reliable.

Data mining is part of a more comprehensive process technically designated as KDD (Knowledge Discovery in Databases) and is not limited to the extraction procedure only: in particular, it implicates also the identification of new, valid schemes leading to useful and proactive actions.

 

Data Mining and Business Intelligence

Although data mining falls into the data science industry, they clearly do notshare the same purpose: the former analyses historical data to describe their behaviour, while the latter harnesses such data bulk to try and predict future schemes.

This feature is quite beneficial in the business intelligence procedures, which by default favour data processing to generate new strategies leading to an improvement of performances and business risk monitoring. Data mining is thus a key component of BI, aimed at obtaining unknown information, schemes or anomalies.

Data mining applications in business intelligence can affect several business areas, especially in the banking sector: from risk monitoring (risk management) to compliance (regulatory compliance) and those business functions responsible for supervisory reporting (primary reporting).

 

Data mining models

There are two types of data mining: descriptive and predictive.

The descriptive model mostly resembles traditional extraction, which recognizes the association between similar data in the starting historical database and regroups the elements presenting analogies.

On the other hand, the predictive model performs an in-depth data analysis aimed at discovering outcomes previously unknown, leading to non-evident knowledge with respect to simple data analysis. This model is more suitable to businesses to identify strategic and operational areas of intervention using the results of the analysis processed by the algorithm, which could not be detected through other methodologies.

Methodologies and scopes of application

Data extraction techniques can differ from one another and, depending on the objectives, different methods could be used. The main ones could be divided into neural networks, decision trees, clustering and Bayesian methods.

These methodological principles are used in several areas where data mining could apply. The various scopes of application particularly includethe economic-financial market, science, marketing, communication technology, statistics and the industrial sector.

 

The advantages of Data Mining

The major advantages obtainable from data mining are quite obvious:

• Quantitative and qualitative analysis of large data warehouses;
• Quick analysis of massive databases;
• Study of different types of data;
• Processing of a large number of variables;
• Clear and simple display of outcomes;
• No a-priori assumption influencing or directing the extraction;
• Use of algorithms dedicated to the function to be carried out.

In a nutshell, data mining is a key resource for businesses and the banking system. It is grounded on a technological infrastructure aimed at algorithm generation, which permeates the whole process.

 

Data Mining in the TIGREARM suite

Data processing and extraction is a central feature of the software suite offered by Save Consulting. The information extraction process to obtain various types of relevant analysis is mainly highlighted in two TIGREARM modules:

• MidaBI: the TIGREARM module dedicated to business intelligence which allows the management of regulatory reporting data base, as well as an in-depth cross-check of data quality (data governance). MidaBI provides for the possibility to generate a wide range of both structured (i.e. pre-defined) and free-format reporting. Free reporting is defined and populated through the production of indicators, value classes and the relative analysis of time mismatches provided by the data mining process.
• Bank of Italy feedback: the TIGREARM module enabling the Business-System comparison, which is one of the key components of Information Technology.

Again, data mining is functional for determining qualitative reference indicators aimed at internal analysis of the data assets, as well as the bank’s positioning degree relative to such dimensions.

 

Discover the TIGREARM modules featuring data mining functions in the dedicated sections and choose the module which suits your needs: MIDABI – BANK OF ITALY FEEDBACK.

TIGREARM can be accessed everywhere, as each module of the suite is web-based