01-09-2014 Outline Preconditions Outcome: What are you supposed to - PDF document

01-09-2014 Outline Preconditions Outcome: What are you supposed to learn? The framework and definition of herd management The management cycle Classical production theory Advanced Quantitative Methods Limitation of classical theories in Herd Management Outline of the course Teachers Course introduction The concept of uncertainty Anders Ringgaard Kristensen Preconditions Brush-up courses … Courses • Mathematics: ”Matematik og The course will start up with brush-up courses of modeller”/”Matematik og planlægning” • Probability calculus and statistics • Statistics ”Statistisk dataanalyse 2” • Linear algebra • Mandatory first year (economics, statistics etc) Learning outcome Outcome - knowledge After attending the course students should be able to After completing the course the student should be able to: participate in the development and evaluation of new tools Describe the methods taught in the course for management and control taking biological variation and Explain the limitations and strengths of the methods in observation uncertainty into account. relation to herd management problems. Give an overview of typical application areas of the methods. 1

01-09-2014 Outcome - skills Outcome: Competencies: After completing the course the student should be able to: After completing the course the student should be able to: Construct models to be used for monitoring and decision Evaluate methods, models and software tools for herd support in animal production at herd level. management. Apply the software tools used in the course. Transfer methods to other herd management problems than those discussed in the course. Interpret results produced by models and software tools. Herd Management Science The management cycle: Classical theories Basic level: (Scarce Utility Resources) Theory, • As we define the basic level, it consists of Ch. 3. • Utility theory • Neo-classical production theory • Basic production monitoring • (Animal nutrition, animal breeding, ethology, Basic farm buildings) Neo-classical Production Production Monitoring, • What any animal scientist should know about Theory, Ch. 5. management Ch. 4. • The starting level of this course • Volume I of the textbook! (Animal science, Production function) Neo-classical production theory How much to produce Answers 3 basic questions: One factor x and one product y • What to produce. Prices p x and p y • How to produce. A production function y = f( x ). • How much to produce. Profit u ( x ) = yp y – xp x = f(x) p y – xp x Marginal considerations Basic principle: Continue as long as the Problem: marginal revenue, MR, exceeds marginal • Find the factor level that maximizes the profit costs, MC. At optimum we have MR = MC. 2

01-09-2014 How much to produce How much to produce Maximum profit where u’ ( x ) = 0. u ( x ) = f(x) p y – xp x 1 u’ ( x ) = f’( x ) p y – p x Total revenue, f( x ) p y 0,8 u’ ( x ) = 0 ⇔ f’( x ) p y = p x 0,6 Maximum profit where: • Marginal revenue = Marginal 0,4 cost! Average revenue, f( x ) p y / x 0,2 0 Marginal revenue, f’( x ) p y -0 ,2 How much to produce, optimum How much to produce, logical bounds 1 1 Total revenue, f( x ) p y Total revenue, f( x ) p y 0,8 0,8 0,6 0,6 0,4 0,4 Average revenue, f( x ) p y / x Average revenue, f( x ) p y / x 0,2 0,2 Price of factor p x 0 0 Marginal revenue, f’( x ) p y Marginal revenue, f’( x ) p y -0 ,2 -0 ,2 Limitations of neo-classical theory Background for course Static approach: Structural development in the sector • Immediate adjustment • Increasing herd sizes • Only one time stage • Decreasing labour input Deterministic approach Technological development • Ignores risk • Sensors, automatic registrations • ”Biological variation” • Computer power • Price uncertainty • Networking Methodological development Knowledge representation (knowledge considered as certain): • Statistical methods • Unobservable traits • Operations Research • ”Production functions” • Detached from production: No information flow from observations. • No updating of knowledge. 3

01-09-2014 Outline of course - I Outline of course - II Part I: Part II: The problems to be solved • Brush-up course on • From registrations to information, value of information, information as a factor, sources of • Probability calculus and statistics information • Linear algebra • Decisions and strategies, definition and knowledge • ”Advanced” topics from statistics foundation • Basic production monitoring • Consequences of decisions and states • Registrations and key figures • Visualisation and user interfaces • Analysis of production results Outline of course - III Teachers Part III: The methods to be used • State of factors Anders Ringgaard Kristensen • Monitoring and data filtering Cécile Cornou • Bayesian networks Dan Børge Jensen • Decision support • Decision graphs • Simulation (Monte Carlo) • Linear programming (low priority) • Markov decision processes (dynamic programming) • Mandatory reports Mandatory reports The web 4 minor reports must be handed in Absalon • Based on the exercises Home page of the course At least 3 must be approved in order to attend the oral • http://www.prodstyr.ihh.kvl.dk/vp/ exam The 4 reports are distributed over the following • Course description methods: • Plan • Bayesian networks • Pages for each lesson with a description of the • Monitoring and data filtering contents, literature, exercises, software to use • Linear programming etc. • Markov decision processes 4

01-09-2014 Exercise, uncertainty Uncertainty, II Production function: Adding uncertainty to production function: • Considerable improvement, BUT = f x x x ( , , ) • Significant uncertainty about true energy, protein 1 2 3 and fat content still ignored c x + c x + c x + c x + c x + c x + c x x + c x x + c x x 2 2 2 11 1 22 2 33 3 1 1 2 2 3 3 12 1 2 13 1 3 23 2 3 • Example, only considering energy • - milk yield given energy, protein and fat Adding uncertainty, the actual milk yield is Y = f ( x 1 , x 2 , x 3 ) + e Uncertainty - III Uncertainty - IV Effects of decisions will be over-estimated if uncertainty about Silage obs.* Silage true Ration Milk yield* • true state • factor characteristics • factor effects Concentr.* Herd size* is ignored. Wrong decisions may be made. True energy content of silage is unknown The precision of the observed content depends heavily on the observation method (standard value from table, laboratory analysis etc.) Uncertainty , V Uncertainty - VI Baysian networks with decisions and Uncertainty is not the opposite of knowledge utilities added (student project). Uncertainty is a property of knowledge Reduction of uncertainty is often possible at some cost! Reducing uncertainty is not always profitable. Silage obs.* Silage true Ration Milk yield* Concentr.* Herd size* Method Price Mix Cost Rev. 5