Telecomunication
The telecommunications industry is one of the fastest developing sectors of the modern economy. The range of services offered by telecommunications companies is continually expanding, while the preferences and behaviour of clients is also constantly changing.

In order to better meet client expectations it is necessary to achieve a deep understanding of their needs. The basic method of finding out about client behaviour is analysis of data collected during day-to-day operations. By using software such as STATISTICA Data Miner telecommunications companies can fully exploit the huge potential of information hidden in their data collections and therefore formulate an effective market strategy.

Here are some of the ways in which STATISTICA Data Miner can be used by telecommunications companies:

Ensuring client loyalty,

Prevention of client loss (churn),

Optimisation of offers and subscription fees,

Market segmentation,

Cross-selling and up-selling campaigns,

Identification of fraud and other irregularities in the use of telecommunications services.

Besides these applications STATISTICA Data Miner also has several technological applications. In particular it can be used for:

Analysis of reliability and damage time of equipment (e.g. transmission elements) which is extremely useful when deciding whether to replace particular pieces of equipment,

Identification of the causes of network malfunctions e.g. establishment of the type of problem and reasons for the disruption of services,

Prevention of client churn.

Client churn is a serious problem for telecommunication companies: not only is the cost of securing clients in this market very high, clients can also easily defect to competitors with minimal cost to themselves. The advanced data mining functions included in the STATISTICA Data Miner system can help identify the reasons for churn and indicate which clients are most likely to defect, thus allowing managers to plan appropriate measures to restrict the scale of the problem. In some cases the telecommunication service provider may itself choose to terminate a subscriber's contract, for example if the client fails to abide by the terms of his contract. STATISTICA Data Miner includes many techniques for building models of this kind e.g. classification trees, neural networks or MARSplines.

Segmentation

STATISTICA Data Miner can also be used to carry out market segmentation i.e. identify groups of clients who are similar to each other. This kind of analysis can help a company optimise its business operations and treat its clients more individually. The system can be equipped with suitable segmentation techniques for analysing data clusters and Kohonen networks and also possesses a wide range of visualisation techniques for describing market segments.

Optimisation of charges

STATISTICA Data Miner is also an invaluable tool for optimising service offers and charges. More information about how to use the programme for this purpose can be found in the case study describing STATISTICA Data Miner applications for the telecommunications industry.

STATISTICA Data Miner - techniques and tools

STATISTICA Data Miner's modelling procedures allow for a wide range of other applications e.g. identification of which clients generate the most revenue, are most loyal, or who are most at risk of bankruptcy. The system uses state of the art linear and non-linear modelling techniques, trees, and neural networks.

STATISTICA Data Miner also includes the General Neural Networks Explorer module which provides the user with tools for classification, forecasting and identification of hidden structures, combined with an automatic designer which allows easy solution of even the most difficult tasks and execution of highly complex analyses.

STATISTICA Data Miner's Feature Selection and Variable Filtering module allows selection of essential features of the items being analysed (e.g. clients, transactions) out of up to a thousand potentially useful variables. The system also incorporates techniques for reducing the number of variables taken into account in modelling, including principal component analysis and the partial least squares method.