Strategies for Solving Large Complex Unstructured Problems

Strategies For Solving Large Complex Unstructured Problems-Free PDF

  • Date:01 May 2020
  • Views:14
  • Downloads:0
  • Pages:27
  • Size:5.56 MB

Share Pdf : Strategies For Solving Large Complex Unstructured Problems

Download and Preview : Strategies For Solving Large Complex Unstructured Problems


Report CopyRight/DMCA Form For : Strategies For Solving Large Complex Unstructured Problems


Transcription:

The challenge of large complex unstructured problems. A financial case study,The statistical engineering approach. The fundamental principles of statistical engineering. The Challenge of Large Complex Unstructured Problems. The vast majority of textbook problems have a single correct. In the words of Meng 2009 they correspond to a recognizable. textbook chapter, However many real problems students eventually face are too. large complex and unstructured to have a correct solution. Khatchatryan 2015 We should enable students to engage in. collaborative teamwork to solve nonstandard and ill defined. problems As in the real world no cookie cutter approach is. sufficient when tackling the business problems inherent in the. proposed case studies as those involve messy data and many. blended intertwined and moving parts,Are we students prepared for the real world. Typical Attributes of Such Problems,Large payoff and business critical problems. Impact is broad process performance financial customer social. Several departments groups and functions are involved. Problem has high degree of complexity involving both technical and. non technical challenges,Problem not clearly defined structured.
There is no known solution,Potential team conflict on how to approach. Multiple sources of data and information are needed. How should practitioners attack such problems,Typical Attributes of Such Problems. More than one technique is required for solution, Typically both statistical and non statistical techniques are also. Creative use of information technology IT is needed. Long term successes requires embedding solution into work. processes typically through,Use of custom software. Integration with other sciences and disciplines,What literature exists to guide practitioners.
Interesting Course Taught at Harvard,Stat 399 Problem Solving in Statistics. emphasizes deep broad and creative statistical, thinking instead of technical problems that correspond. to a recognizable textbook chapter,Xiao Li Meng American Statistician August 2009. Do large complex unstructured problems,correspond to a recognizable textbook chapter. Stat 399 Raises Important Questions,What approaches should be used to attack large.
complex unstructured problems, By what theory or body of research should we answer. this question,How would we know if we were right or not. Should this be a discipline in its own right,A Personal Case Study. Problem GE Capital announced losses of over 125 million. on WorldCom bonds that went into default, Their question to GE Global Research Is this just the cost of. doing business in the financial sector or could we have. predicted these losses with enough lead time to lower our. Classic problem in finance unsolved,A Personal Case Study.
Challenges included, Financial theory efficient market hypothesis says you can t. predict defaults ahead of the market,GE Capital needed to trade in large quantities. No commonly accepted definition of default,Limited internal data no set of universal data. No defined measure of success,Does this sound like a typical textbook problem. A Personal Case Study,Approach Taken,Cross functional team organized.
Statistics operations research machine learning quantitative. finance business expertise, Spread between upstate New York Bangalore India and. Stamford Connecticut, Developed definition of default and metrics to document. success and failure step zero,Typical for unstructured problems. No template to follow,A Personal Case Study, Data obtained externally needed to merge different data. Eventually set up direct data feed from Wall Street. Final prediction methodology utilized,Publicly available default predictor as an input.
Engineering vs pure science approach, Smoothing algorithms classification and regression trees. CART simulation and Markov Chains, Developed control plan to detect need for retuning the. algorithm used censored data methods from reliability. Does this look like a typical textbook solution,A Personal Case Study. GE Capital performed a simulation study of the final prediction system. without our involvement, Evaluated their potential financial results had they used this system in the. past year for all trading, Results were positive in the hundreds of millions of dollars.
This system was subsequently incorporated into underwriting. procedures for large financial deals, Embedding statistical methods into business processes. The team received a patent for the system not for the algorithm. US20030229556A1, Solving large complex unstructured problems produces impact. The Statistical Engineering Approach, How was this problem attacked Answer using what we. now call statistical engineering,Definition, The discipline of statistical engineering is the study of the systematic. integration of statistical concepts methods and tools often with other. relevant disciplines to solve important problems sustainably. In other words trying to build something meaningful from. the statistical science parts list of tools,Focus is on solving problems versus tools per se.
Real problems particularly big problems require integration of multiple. See special edition of Quality Engineering 2012 on statistical. engineering for more background and case studies,Statistical engineering is not a method per se. Key Aspects of Definition,the study of,Research oriented. Statistical engineering has a theory,solve important problems sustainably. Results are the what methods and tools are hows, Statistical engineering is therefore tool agnostic. Solution must be sustainable over time,often with other relevant disciplines.
Integration of multiple tools methods and even disciplines. Information technology usually has a major role to play. SE studies how to select and integrate methods in order to solve real problems. Solving large complex unstructured problems produces impact 14 The Statistical Engineering Approach How was this problem attacked Answer using what we now call statistical engineering Definition The discipline of statistical engineering is the study of the systematic integration of statistical concepts methods and tools often with other relevant disciplines to solve important

Related Books