Data Mining in Airlines - Literature review Example

Data mining in Airlines Lecturer: Efficient Computer Experiment-Based Optimization through Variable Selection SectionA The study evaluates variable selection for use in regression in data mining through designed experiments as well as statistical modeling useful in representing complex objective function evaluated only by solving optimization sub problem (Shih et al., 2014). Since huge applications have a huge number of variables, direct utilization of computer experiments requires exceedingly large designed experiment as well as considerable computational effort. Naturally, variable selection can be conducted after executing small computer experiments. However, conventional techniques of variable selection cannot be applied in huge variables. Therefore, the study explores utilization of regression trees as well as multiple testing procedures where performance of chosen variables evaluated by determination coefficient as well as relative errors (Shih et al., 2014). There are classifications and regression trees (CART) are the common data mining techniques in supervised learning. CART algorithm utilizes binary recursive screening to divide variable space to rectangular regions in accordance with the similarity within the values of response. The commonly used multiple testing is false discovery rate (FDR) procedure. FDR procedure used in variable selection necessitates a definite response variable, which separates data into groups (Shih et al., 2014). Section B Variable selection in the study is significant when considering a large number of explanatory candidate variables, which also believed to contain many unnecessary or redundant variables. The conventional two-phase stochastic programming utilizes Bender’s approach. However, two-phase stochastic programming in large-scale problems tends to be sluggish to converge, hence use of two-phase DACE (design and analysis of computer experiments) approach decreases computation used in carrying out optimization (Shih et al., 2014). The solution the study provides is variable selection that speeds up large-scale optimization techniques according to the design as well as analysis of computer experiments. In the study, DACE is a design utilized in organization of a collection of executed computer experiments in order to facilitate appropriation of statistical “metamodel” which approximates measure of performance from the computer experiment (Shih et al., 2014). For an existing airline network owning 50 stations with 2358 legs, DACE Phase decreased decision parameters to 1264 dimensions from 6537 with multi-step procedure deriving 141 preliminary extreme points that were later extrapolated to 3562 design points within the practical region. Then the second-phase subproblem is solved for every one of the design points. Of the 1264 variables for decision, many unnecessary ones still exist that can be established through variable selection, to provide a considerably small collection of design points (Shih et al., 2014). Section C The use of subset in the study maximizes or minimizes the appropriate criterion. In the study, two obvious subsets include the entire set of variable or the finest single variable. However, problem lies in the selection of intermediate subset, which is superior to the two extremes. The weakness lies in finding the necessary variables among the complete collection of variables. A novel decision rules approach for customer relationship management of the airline market Section A The topic is about customer behavior and firms developing permanent relationships with customers (Liou, 2009). The problem the study tries to solve is precisely forecasting customer behavior to help firms minimize loss of existing customers by vigorously creating an enduring relationship with customers (Liou, 2009). The paper uses factor analysis; as well, it introduces Variable Consistency Dominance-based Rough Set Approach (VC-DRSA) within customer relationship management (CRM) in airline market (Liou, 2009). Section B Literature regarding data mining as well as its application in CRM in the airline market is virtually silent. Various firms ought to be capable of determining their customers’ significance in order to keep or even nurturing potential profitable customers. Customer relationship management (CRM) remains an essential part in the current airline business owing to globalization, market saturation, enhanced competition, and swift technology advances. CRM aim involves understanding the profitability associated with customers and retaining profitable ones (Liou, 2009). CRM being a dynamic procedure for running a customer–company relation ensures that customers choose to prolong mutually valuable commercial exchanges while restraining themselves from participation in exchanges, which are unprofitable to the company. CRM remains an essential business strategy that a firm should focus on desires of its customers; as well, it must integrate customer-oriented methodology all over the organization (Liou, 2009). The paper applied the VC-DRSA in CRM of airline market, which is offered in the paper. Contrary to traditional statistical techniques like analysis of discriminant, the potency of rough set theory is it never requires fundamental statistical assumptions (Liou, 2009). Section C In several real-life issues, organizing properties of the measured attributes have a significant role. For example, such characteristics of objects like product quality and market share often receive typical consideration as criteria in economical problems (Liou, 2009). Rogue components: their effect and control using logical analysis of data Section A The study is about rogue components that plague the airline business by creating havoc in programs of asset management. The study describes the way rogues components develop, summarizes the process leading to hampering asset management program; as well, it examines various negative effects that ensue (Mortada et al., 2012). The study main problem is controlling the development of rogue components. The study uses supervised learning technique of data mining known as Logical analysis of data (LAD) within CBM with the aim of detecting rogue elements within a collection of repairable elements. The study applies the resultant LAD decision model on a range of turbo compressors owned by an airline fleet (Mortada et al., 2012). Section B Maintenance as well as reliability programs in aircrafts are essential in guaranteeing wellbeing and airworthiness associated with airplanes. Rogue components are very difficult to recognize and is capable of spreading throughout the component population. The major problem surface when rogue elements find their way to asset management programs through spare part inventory for operator. The discovery of such elements is essential to guarantee reliability. The rationale for a component developing rogue breakdown arises from repair or even overhaul tests that never address 100% the component’s functioning, features or environment (Mortada et al., 2012). In CBM, detection of fault is only achieved where there are indicators capable of revealing information regarding the asset status by screening them. The skill to exploit these indicators relies on where, within the maintenance procedure, rogue element detection occurs. Implementing LAD algorithm can occur at one point in the process either before or after the component gets stored in the repair shop. Carrying out detection before repair, any needless resources, which may be spent on a rogue elements are saved (Mortada et al., 2012). Results of the tests in the study illustrated LAD technique has the ability to detect automatically rogue components by feeding performance history of the components into the LAD algorithm. Automatic rogue components detection solves the issue of sifting through records to visually evaluate each component (Mortada et al., 2012). Section C The benefit of using rogue component detection is that it saves a lot of time as well as resources since LAD can accomplish in seconds a task that currently takes days in the industry. However, performing detection of rogue components before repair means that LAD algorithm relies on other indicators in generating decision regarding rogueness in a component (Mortada et al., 2012). Security investment in aviation industry: a longitudinal analysis Section A The study evaluates the way environmental transformation and project features affect the distribution of security-associated Airport Improvement Program (AIP) funding with aviation industry (Wang, Andoh-Baidoo & Sun, 2014). Attention toward secure transportation dates back to a succession of terror attacks carried out on surface and air-borne vehicles. Terrorist attacks on transportation motor vehicle or even a facility may produce disastrous personal injury, enormous loss of assets, and vast effects to public psychology. One significant function associated with government agencies involves providing financial as well as technical support toward transport carriers in order to improve security procedures (Wang, Andoh-Baidoo & Sun, 2014). The study evaluates longitudinal trend regarding federal security investment in transportation and its cross-sectional components (Wang, Andoh-Baidoo & Sun, 2014). Section B Among the diverse transportation security parts, airport security remains the most contentious because it is vastly invested however it raises concerns. Before the 2001 September 11 terrorist attack, Federal Aviation Administration (FAA) had the responsibility of guaranteeing airport security with the major source of security funding being Airport Improvement Program (AIP). However, following the attack, Congress Act approved by the then President Bush allowed creation of TSA and tasked it with most of “land-side” security responsibility. In spite of the changes regarding security-related tasks from FAA to TSA, still FAA is responsible to a section of security in airports, particularly the land-side functions (Wang, Andoh-Baidoo & Sun, 2014). A conceptual connection to the word “security” divulges several aspects of AIP security initiatives supported through government funding. Results of the study supports theoretical framework that two levels exist that influence the distribution of grants in transportation security and they include environmental changes in macro level as well as project features in micro level. Moreover, analysis points out that 9/11 events and the economic crisis in 2008 greatly affected the provision of grants associated with security and all grants in various ways (Wang, Andoh-Baidoo & Sun, 2014). Section C The main weakness in the analysis of the study is that it is restrained by the variable choices because of secondary data. For instance, airport region and type are the only two variables describing project characteristics that only approximately indicate location and scope. Moreover, FAA discloses only financed AIP projects, increasing the difficulty of investigating factors that influence awarding of grants. Despite the weaknesses, the study findings still generate some significant implications to both practitioners and researchers. Moreover, it shows that investment in transportation security positively responds to terrorist attacks like 9/11 although this is not the case with economic events like the 2008 economic crisis (Wang, Andoh-Baidoo & Sun, 2014). List of references Liou, J. J. (2009). A novel decision rules approach for customer relationship management of the airline market. Expert Systems with Applications, 36(3), 4374-4381. Mortada, M. A., Carroll Iii, T., Yacout, S., & Lakis, A. (2012). Rogue components: their effect and control using logical analysis of data. Journal of Intelligent Manufacturing, 23(2), 289-302. Shih, D. T., Kim, S. B., Chen, V. C., Rosenberger, J. M., & Pilla, V. L. (2014). Efficient computer experiment-based optimization through variable selection. Annals of Operations Research, 216(1), 287-305. Wang, Y., Andoh-Baidoo, F. K., & Sun, J. (2014). Security investment in aviation industry: a longitudinal analysis. Industrial Management & Data Systems, 114(2), 276-291. Read More