Customer targeting is the key to the success of bank telemarketing. To compare the flexible discriminant analysis and the logistic regression in customer targeting, a survey dataset from a Portuguese bank was used. For the flexible discriminant analysis model, the backward elimination of explanatory variables was used with several rounds of manual re-defining of dummy variables. For the logistic regression model, the automatic stepwise selection was performed to decide which explanatory variables should be left in the final model. Ten-fold stratified cross validation was performed to estimate the model parameters and accuracies. Although employing different sets of explanatory variables, the flexible discriminant analysis model and the logistic regression model show equally satisfactory performances in customer classification based on the areas under the receiver operating characteristic curves. Focusing on the predicted “right” customers, the logistic regression model shows slightly better classification and higher overall correct prediction rate.

This thesis provides analytical and forensic tools for data cabling, with particular focus on Ethernet cabling to assist designers and those involved in deployments in analyzing cable performance and the reasons behind the actual performance obtained. The need for higher bandwidth to accommodate increasing demand for multimedia services and data centers network infrastructure led to the formation of IEEE P802.3bq to create standards for 40GBASE-T over twisted pair cables. The 40GBASE-T is expected to offer bandwidth of up to 2000MHz over a maximum channel length of 30m. The research investigated means of predicting key performance parameters in Ethernet cabling standardization using the 40GBASE-T as an example. The performance parameters prediction method provided is equally applicable to ongoing and future high data rate Ethernet cabling standardization such as the 2.5/5GBASE-T and 50/100GBASE-T. Another problem in the Ethernet networking world is the availability of counterfeit and non-standards compliant twisted pair cables in the market. The significant amount of communications cables in the market containing copper clad aluminum cable or other non-standards compliant conductors disguised as Category 6 cables can pose serious problems to companies’ networks, the contractors or the installers. This is in view of the growing demand for internet of things (IOT) services that makes it imperative to have a reliable Ethernet driven communication network to support the required infrastructure. This thesis therefore, provides techniques that can be used to evaluate cables key performance parameters using the Feature Selective Validation method and the Kolmogorov-Smirnov (KS) test. The technique can help engineers avoid subjective judgement and make objective decisions in the selection of cables. The research provided a technique that can be used to reverse engineer impedance profile from the return loss measurement of Ethernet cables using genetic algorithms. The method can be applied in situations where time domain tests are inaccessible or only simple (magnitude) tests in the frequency domain are available and there is the need for impedance profiles of cables to evaluate their performance or physical integrity before or after installation. The method can also be useful where only simple (magnitude) tests are the only historical data available for the cables and facilities for time domain reflectometry measurements are inaccessible. This research also presented a method of evaluating and predicting NEXT in unshielded twisted pair (UTP) using Category 6 cables as an example. The results obtained from the evaluation were used to provide crosstalk parameters for fast NEXT prediction in Category 6 (UTP) cables. The research used the measured NEXT of three Category 6 (UTP) cables from different manufacturers for evaluation and validation. The evaluation and modeling method can thus be useful to engineers investigating NEXT in the design of data communication systems.