Incident Reporting System in Airline Business - Essay Example

Incident Reporting System in the Airline Business Aviation safety is important for airlines because accidents can be expensive and can cause major operational upsets as well as raising questions about the image of an airline in the minds of the passengers, who generate an income for the airline. Hence, there is a need for an incident reporting system that will log safety related reports and issues which can be analysed in order to generate information which will be of use in preventing accidents or safety threats in airline operations. Because safety in airline operations is important, therefore such a system must have the support of the highest levels of management. Most airlines do have a safety related incident logging system and in the United States of America, the Federal Aviation Authority or the FAA also maintains such a system. The airline safety incident logging system maintains a database of all safety related reports or issues which have been made available to the airline and these can then be analysed using data mining or analysis techniques such as trend analysis, text mining, clustering and statistical analysis to generate valuable information about the nature of safety related issues which are on the rise and may cause a problem. Airline management can then take corrective action to improve the situation and reduce the level of incidents related to particular areas of operations. Using incidents logged per 100,000 departures is a common measure for aviation safety. Relational databases, data mining tools, free text retrieval and probabilistic inference techniques are used as the data stored in the incident logging system grows and changes over a period of time. Complex query formation for the database may be assisted by techniques such as conversational case based reasoning which places a greater emphasise on certain aspects of a query . This brief essay takes a look at the architectures and analysis techniques for airline safety incident logging systems. Introduction Because the consequences of a lapse in aviation safety can turn into a major catastrophe and have such a major impact on the earnings, image as well as the operations of an airline, there is a requirement for the responsibility for aviation safety to be taken seriously by the senior managers of an airline. Through its attitudes and actions, senior management influences the attitudes of others and impresses upon all the other members of an airline staff that aviation safety has to be taken seriously. Although issues related to safety are not the primary business of an airline, safety is important because it is a supporting issue (Flight Safety Australia, 2001, Pp 33 – 35). Even though managers may be tempted to allocate more resources to the production related processes in an airline, issues related to safety have to be adequately covered and all airline personnel have a contribution to make in order to ensure that incidents which can have an impact on airline safety are brought to the attention of managers for analysis and remedial action. For each accident that occurs, there are hundreds of incidents spread over time which may have assisted in the identification of a problem which had an association with the root cause of the accident. It is for this reason that most airlines have an incident reporting system in place which serves to log incidents and analyses these incidents to present useful reports to senior managers so that remedial action can be taken (Flight Safety Australia, 2002, Pp 23 – 28). In the United States of America, the Federal Aviation Authority or the FAA maintains an aviation incident reporting system and airlines maintain their own systems for the analysis of incidents which may have a bearing on aviation safety. Because the incident related data can grow over time, therefore it is required to maintain such incident data in a relational database, which can also make it possible for such data to be analysed using statistical analysis software or data mining tools as well as trend analysis. The results of such analysis can then be condensed into reports which influence the remedial action that must be taken. A holistic approach built into intervention strategies is normally required (Johnson, 2000, Pp 1). Within an airline, the generation of an adverse occurrence will result in the generation of a report by the contributors, who are usually the airline staff who may have witnessed the adverse occurrence. This report is submitted to the delegated authority within the airline organisation and the process of supplementing as well as validating this incident is undertaken by the delegated authority. Corrective action may be required to be undertaken as a result of the analysis of the incident and the incident is added to an incident database. If a particular incidence is important enough to be mentioned to relevant senior managers, then a summery of such an incident is immediately communicated, otherwise the senior managers are kept informed through reports that are generated as a result of the analysis of incidents over a period of time (Johnson, 2001, Pp 1 – 10). Trend analysis is one of the simplest techniques which can be used to analyse incidents of a particular type related to airline operations. A rising trend in incidents of a particular type will indicate that there is a problem associated with some processes or a particular area of airline operations. Corrective action will have to be taken immediately in response to a rising trend of incidents of a particular type. However, in large airlines, or for incident logging systems involving a number of airlines, the number of incidents that are being logged can become large. FAA, for instance, uses incidents logged over 100, 000 departures to determine if a satisfactory record of the level of incidents is being maintained. Some incidents which are logged can also of a complex nature, which require sophisticated analysis from various perspectives. Hence, sophisticated statistical analysis or data mining techniques may also be required to be employed in order to present comprehensive reports which can be of value in enhancing safety (Rothleder, 1998, Pp 2 – 8). This brief essay takes a look at some of the issues associated with the use of trend analysis, software tools and techniques for the analysis of incident reporting system data in an airline. Computational Techniques for Incident Reporting in the Airline Business Many incident reports are textual in nature and appropriate forms have to be designed in order to ensure that the relevant information which is required for incidence analysis is passed on to the delegated authority handling incidence reporting. The incident reports have to be correctly indexed and classified and because an airline operates over a very distributed geographic region of operation, the analysis of incidence reports can be a problem. Some classifications which have been used in the airline include weather related problems, problematic ground conditions, pilot errors, cabin crew errors, equipment malfunction, maintenance errors, airline procedure error, air traffic control error, passenger error, birds and pilot error of other airplane etc (Haunschild, 2002, Pp 9). Because of the number of incidence reports that have to be handled, relational databases are the most appropriate means for the storage of incidence related reports. Manual classification of these reports is difficult because of the number of incidents. However, at some stage in the processing of the incidence reports, a suitably trained and experienced team of humans will consider these reports. Because of the volume of incidence reports that are likely to be generated, it is highly desirable that text mining tools which are available for reputed relational database packages such as the IBM DB2 or the Oracle 9i – 11g type databases be used to automate the process of classification and indexing as well as analysis. The previously mentioned databases are highly scalable and supported by many advanced analytical tools which can assist in an analysis of the data in the airline incident reporting system. Once the process of classification and indexing of the incidence reports has been accomplished, it is appropriate to analyse the whole database for trends in certain classifications of incidents. Analysis tools which are available with the previously mentioned database packages can not only perform trend analysis, but also other statistical analysis such as trend analysis or clustering analysis (Johnson, 2000, Pp 1) and (Carbone, 2000, Pp 2 – 5). A rising trend of incidents of a certain type will indicate that there is a problem in the processes or airline operational areas associated with the classification. Attempts can be made to explore if more sophisticated correlations between variables or incidents can be filtered out from the data. Some large airline or aviation incidence reporting systems can have as many as 500,000 reports which have been accumulated over time and hence query generation for the database can be difficult. Systems which use probabilistic information retrieval can be attempted to be included in the software tools or built into the query system for such incident reporting systems which emphasise on certain characteristics when searching through the stored incident reports. The analysis system can be automated and programmed to identify urgent potential hazards or recognise precursors which may be required to be investigated further (Tamuz, 2000, Pp 2 – 10). Aviation safety incidents may be broadly classified as being related to mechanical causes or human error. Many further sub-classifications of mechanical incidents related to safety are possible. An analysis of the data stored on the incident reporting system will enable managers to develop remedial programs. An obvious remedial program for human errors is further training in procedures or in the handling of situations that are likely to occur. It is also possible to modify certain business processes so that they are more user-friendly and hence less prone to human errors. Remedial action for incidents related to what may be classified as mechanical incidents will usually require engineering investigations that may be required to be very thorough. Even design faults in new models of aircrafts have been discovered as a result of an analysis of the so called mechanical related incidents. However, even mechanical safety incidents are most often traced back to some errors in a methodology associated with aircraft operations, maintenance or a process. Incidents related to human error can be further traced back to cognitive, aero medical, psychosocial, organisational or ergonomic factors. The important thing is to realise that attempts have to be made to consider and analyse the data in the airline incident reporting system carefully and from a number of perspectives. It helps if the personnel associated with the analysis of the data on the incident reporting system be highly experienced and well trained. Maintenance errors which can also be considered to be human errors are an important dimension in safety related incidents. Airlines and staff generally learn from the knowledge that is generated as a result of the analysis of incident reports in the airline incident reporting system (Taneja, 2002, Pp 2 – 5) and (Haunschield, 2002, Pp 4). For the relatively small airlines with a few aircrafts, the numbers of incident reports that are generated are likely to be small and hence adequate results can be generated from the relatively unsophisticated software tools. However, for the larger airlines with a substantial and varied fleet in service, the complexity as well as heterogeneity of the incidents will escalate. Although many industries maintain even larger databases, the precision and recall of safety incident reports for a large airline can become important for analysis. High recalls but low precision in response to a database query can mean that a large number of relevant documents for the query can be retrieved along with a relatively large number of irrelevant documents. There can also be problems with data abstractions and dynamic classifications as to the nature of the safety related incidents changes over time and new classification schemes are used to classify as well as index incident reports. As an example, a query about “turbine blade problems” will fail if all the incident reports have been classified as “engine problems”. Hence, appropriate care has to be taken to ensure that the software tools which are selected for incident reporting system database analysis, the database itself and the analysts investigating the data are appropriate, of high quality and trained in the peculiarities of the reporting system. Hence any human interfaces to the database for querying text searches and selection criteria for the database query inclusion can be made more precise as a result of incorporating a probabilistic inference system for full text query retrieval (Johnson, 2002, Pp 1). In the analysis of data on the airline incident reporting system, any query formation for analysis can often be less sophisticated as compared to the skills of the experienced personnel that have been associated with the analysis of such data for a long time. For large airline incident reporting systems, it is appropriate to try and use tools or to develop interfaces which have case – based reasoning logic built into the software. Case - based logic attempts to determine if some terms in a query are relevant to a query or not using principles of artificial intelligence. The parsers in the to the incident reporting system database can be enhanced with regard to their ability to recognise grammar in order to assist with better processing of queries and analysis (Dragunsky, 1991, Pp 1 – 27). A routine analysis of accident precursors of which trend analysis is also an example can be useful in the analysis of incident reports which have been classified. Such efforts can, therefore, produce valuable insights for the safety management of airline operations. Conclusion A database consisting of incident reports associated with airline operations has been proven to be invaluable in enhancing the safety of airline operations. Hence, such an incident reporting system should be incorporated as a management information system and senior managers should incorporate a culture of openness and the reporting of incidents. An emphasis on safety is important because of the catastrophic financial losses and the losses in terms of valuable human lives which can occur as a result of a mishap. Simple descriptive statistics and software analytic tools for trend analysis cam shed a light on the current state of the safety status of planes and operations. However, new approaches text analysis and anomaly detection as well as sophisticated data mining approaches can be developed for trying to identify more complicated and heterogeneous trends in the incident reports. References / Bibliography 1. Bielefeld University. (2005). The Causal Systems Analysis and WB-Analysis Home Page. Bielefeld University. Retrieved: September 5, 2005. From: http://www.rvs.uni-bielefeld.de/research/WBA/ 2. Carbone, Patricia. Bhavani Thuraisingham and Chris Clifton. (August, 2000). Data Mining. The Edge – MITRE Advanced Technology Newsletter. Retrieved: September 5, 2005. From: http://www.mitre.org/news/the_edge/august_00/8_00.pdf 3. Dragunsky, Cesar. (1991). Analysis of Finite State Grammar for Parsing Airline Safety Reports. Simon Fraser University. Retrieved: September 5, 2005. From: http://ir.lib.sfu.ca/retrieve/162/etd0006.pdf 4. Flight Safety Australia. (March – April, 2002). Special Issue: Safety Management – Making it fit. Flight Safety Australia. Retrieved: September 5, 2005. From: http://www.casa.gov.au/fsa/2002/mar/22_28.pdf 5. Flight Safety Australia. (May – June, 2001). Risky Business. Flight Safety Australia. Retrieved: September 5, 2005. From: http://www.casa.gov.au/fsa/2001/may/32-35.pdf 6. Haunschild, Pamela R. and Bilian Ni Sullivan. (December, 2002). Learning from complexity: effects of prior accidents and incidents on airlines learning. Administrative Science Quarterly. Retrieved: September 5, 2005. From: http://www.findarticles.com/p/articles/mi_m4035/is_4_47/ai_107762238/pg_1 7. Johnson, Chris. (2000). A Brief Overview of Computational Techniques for Incident Reporting. University of Glasgow. Retrieved: September 5, 2005. From: http://www.dcs.gla.ac.uk/~johnson/papers/incident_overview/ 8. Johnson, Chris. (2000). Using Case-Based Reasoning to Support the Indexing and Retrieval of Incident Reports. University of Glasgow. Retrieved: September 5, 2005. From: http://www.dcs.gla.ac.uk/~johnson/papers/case_based_reasoning/ 9. Johnson, Chris. (2001).Architectures for Incident Reporting. University of Glasgow. Retrieved: September 5, 2005. From: http://www.irit.fr/recherches/LIIHS/palanque/WSSUCA2000/suca-Johnson.PDF 10. Johnson, Chris. (2002). Software Support for Incident Reporting Systems in Safety-Critical Applications. University of Glasgow. Retrieved: September 5, 2005. From: http://www.dcs.gla.ac.uk/~johnson/papers/incident_software/ 11. Proefschrieft. (2000). Organised Learning from Small – Scale Incidents. University of Delft. Retrieved: September 5, 2005. From: http://www.library.tudelft.nl/delftdiss/pdf/2000/tpm_koornneef_20000926.pdf 12. Rothleder, Neal J and Earl Harris, Eric Bloedorn. (1998). FOCUSING ON THE DATA IN DATA MINING: LESSONS FROM RECENT EXPERIENCE. The MITRE Corporation. Retrieved: September 5, 2005. From: http://www.mitre.org/work/tech_papers/tech_papers_97/afc/data_mining.pdf 13. Tamuz, Michael. (2000). Understanding Accident Precursors. University of Tennessee. Retrieved: September 5, 2005. From: http://www.nae.edu/NAE/engecocom.nsf/0754c87f163f599e85256cca00588f49/85256cfb004759c185256dd60053b88d/$FILE/Tamuz.PDF 14. Taneja, Narinder. (2002). Human Factors in Aircraft Accidents: A Holistic Approach to Intervention Strategies. University of Illinois. Retrieved: September 5, 2005. From: http://www.humanfactors.uiuc.edu/Reports&PapersPDFs/humfac02/tanejahf02.pdf Read More