The Joy of Stats: Hans Rosling

this is certainly something worth watching.



and this is a sample of what will be presented today at 21:00 at BBC Four



Vlasios

New Article: Towards a unifying formalisation of geographic representation: the object-field model with uncertainty and semantics

Abstract: The need for a conceptually unifying data model for the representation of geographic phenomena is widely understood. Although some successes have been reported, progress has been slow, especially at the conceptual and logical levels of abstraction. Drawing on and combining existing successes, this article suggests the object-field model with uncertainty and semantics at the conceptual and logical levels of abstraction. The logical level has been formalised in the Unified Modelling Language (UML) class diagram. It is shown that the concepts required to better represent geographic phenomena can be derived from a single foundation that is termed the Elementary_geoParticle with associated uncertainty and semantics by means of aggregation. The town centre phenomenon is used as an application of the conceptual framework being proposed.

To download the paper: http://www.informaworld.com/smpp/content~db=all~content=a930247080~tab=content

Video of oil production

The data has been generated in ACEGES and animated in R.

ACEGES Software: The Interaface

Below is the latest interface of the ACEGES 1.1 software, with the emphasis on the development of exploratory scenarios for oil and gas production.

Screen 1


Screen 2

Benoit Mandelbrot 1924-2010

Benoit Mandelbrot, the mathematician, the father of fractal mathematics, and advocate of more sophisticated modelling in quantitative economics and finance, died on 14th October 2010 aged 85.



For a compilation of videos see: Benoit Mandelbrot

Sustainability and Business

Although the video is not directly related with the aims of the blog, I really like it. That is why it is here.




Maybe, novel modelling tools can help us see things differently.

PhD Opportunities at LondonMet

There are a number of PhD opportunities at LondonMet. Please see here: http://www.londonmet.ac.uk/londonmet/research/the-graduate-school/vice-chancellors-phd-scholarships.cfm

In particular, see the following proposals ( http://www.gamlss.com/papers/PHDoportunities.pdf) :

i) Data Mining Models for the Flexible Modelling of the Location, Scale and Shape Parameters of a Response Distribution
ii) Comparison, Evaluation and Development of Stochastic Volatility Models
iii) Energy-Economy-Investment Modelling
iv) Individual decision-making and social interaction at the movies: The Philadelphia Story

Fusing the agent-based and Object – Field models

One of my recent works appeared as an advance online publication at the Environment and Planning B: Planning and Design. The url of the paper is: http://www.envplan.com/abstract.cgi?id=b36001

Abstract. The fusion of agent-based and geospatial models represents an exciting new synthesis for social science and economics. It has the potential to improve the theory and the practice of modelling complex real-world phenomena. Yet, to date, there has been little systematic analysis at the conceptual and logical levels of how to fuse agent-based and geospatial models for the representation and reasoning of socioeconomic phenomena. Here both sets of issues are explored. In particular, it will be argued that the development of synthetic models requires autonomous agents and flexible organisational structures that can complete their objectives while situated in a dynamic and uncertain geoenvironment represented by the concept of Elementary_geoParticle. As an example of the concept, I present a preliminary conceptual model of global energy to demonstrate the validity and possible uses of the proposed technique.

The modelling framework discussed above has been used in the ACEGES project discussed here: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1652361. A new paper is currently under review in an energy journal (i will not name the specific journal until the review process completes).

If you are interested in the topic of agent-based and geospatial models, see the gisagents blog ( http://gisagents.blogspot.com/ ) by Dr Andrew Crooks.

Simulated Scenarios of Convetional Oil Production

I will present a paper at the 4th International Conference on Computational and Financial Econometrics hosted by University of London & London School of Economics. See here for details: http://www.cfe-csda.org/cfe10/ The paper may also appear at the Computational Statistics & Data Analysis journal.


Abstract:
The ACEGES (Agent-based Computational Economics of the Global Energy System) model is an agent-based model of conventional oil production. The model accounts for four key uncertainties, namely Estimated Ultimate Recovery (EUR), estimated growth in oil demand, estimated growth in oil production and assumed peak/decline point. This work provides an overview of the ACEGES model capabilities and an example of how it can be used for long-term scenarios of conventional oil production. Because the ACEGES model has been developed using the Agent-based Computational Economics (ACE) modelling paradigm, the macro-phenomenon of interest (world oil production) grows from sets of micro-foundations (country-specific decision of oil production). The simulated data is analyzed in GAMLSS (Generalised Additive Models for Location Scale and Shape). GAMLSS is a general framework of modelling where the response variable (oil production) can have a very general (up to four parameters) distribution and all of the parameters of the distributions are modelled as linear or smooth function of the explanatory variable (e.g., time). From a methodological perspective, ACEGES and GAMLSS are applied to help leaders in government, business and civil society better understand the challenging outlook for energy through controlled computational experiments.

Probabilistic forecasts of oil production

Following from my last posting, see the two "improved" figures (analysis conducted with Mikis Stasinopoulos and Bob Rigby).

The H-H scenario:


The L-M scenario:

L-M and H-H Scenarios of oil forecasting

I have created two scenarios of conventional oil production and below you can see the simulated results and the the historical production from 2000 to 2009 (black dots). Given the four key uncertainties (demand growth, production growth, EUR and peak/decline point):

1)The L-M case uses estimates from the demand growth, production growth, EUR and the Monte Carlo process for the peak/decline point (this is the low-medium heterogeneity)

2) All key uncertainties are drawn from the Monte Carlo process (this is the high heterogeneity case)


Note: the simulation is initialized with the 2001 data. Thus, the first simulated year is 2002.

Benoit Mandelbrot: Fractals and the art of roughness

I hope you will enjoy the video as I did:

Presentation of ACEGES at the CEF 2010

This is the presentation that I have at the Society for Computational Economics, 16th International Conference on Computing in Economics and Finance, London, UK.

The ACEGES 1.0 Documentation

We have recently completed the first draft of The ACEGES 1.0 Documentation: Simulated Scenarios of Conventional Oil Production.

Abstract: The ACEGES model is an agent-based model of conventional oil production for 93 countries. The model accounts for four key uncertainties, namely Estimated Ultimate Recovery (EUR), estimated growth in oil demand, estimated growth in oil production and assumed peak/decline point. This documentation provides an overview of the ACEGES model capabilities and an example of how it can be used for long-term (discrete and continuous) scenarios of conventional oil production.

The documentation can be accessed from: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1652361

Citation:
Voudouris, V. and Di Maio, C. (2010), The ACEGES 1.0 Documentation: Simulated Scenarios of Conventional Oil Production, Centre for International Business and Sustainability (London Metropolitan Business School): Working Paper 12, London, UK.

VC PhD Scholarships: ACEGES project

Proposal 1- Energy-Economy-Investment Modelling: Volatility of oil prices and oil supply
Applications are invited in the area of Energy-Economy-Investment Modelling. The applications should be within the scope of the ACEGES project (http://www.londonmet.ac.uk/lmbs/research/cibs/cibs-scenario-planning/cibs-scenario-planning_home.cfm).

In particular, we encourage applications that aim to investigate the relationships between:
i) Spot price of crude oil,
ii) Expectations of future oil prices
iii) Price of crude oil futures
iv) Oil futures spread (defined as the percent deviation of the oil futures price from the spot price of oil)
v) Oil supply shocks.

The overall research aim is to investigate what determines the spot and futures price of crude oil and the importance of the evolution of the price of oil in explaining oil production of OPEC and non-OPEC countries.

Applicants are expected to have a Master’s degree with a strong quantitative, mathematical or statistical focus. Programming experience in Java and/or R is desirable but not essential.

References:
1) Hamilton, J. (2003), What is an Oil Shock? Journal of Econometrics, vol. 113(2), 363-398
2) Alquist, R and Kilian, L. (2010), What Do We Learn from the Price of Crude Oil Futures? Journal of Applied Econometrics, 25(4), 539-573.
3) Kaufmann, R.K. (1991), Oil production in the lower 48 states: Reconciling curve fitting and econometric models. Resources and Energy, 13(1), 111-127

In the first instance, contact me at v.voudouris@londonmet.ac.uk

Proposal 2 - Consumer decision-making and social interaction at the movies: The Philadelphia Story
When faced with choices between new products and services, the properties of which are uncertain, how do individuals behave? Economists commonly distinguish between social and private information available to the consumer - social learning takes place through social networks, while private learning is the consequence of past experience. In factoring both into the decision-making process it is possible to model consumer behaviour - in this instance film consumer behaviour.

To do this I propose developing an agent-based model, which allows the researcher to investigate the microscopic behaviour of individual film consumers who watched particular films at particular cinemas at particular moments in time in order to a) grow the macroscopic environment manifest in the long tail distribution of revenue, b) chart the pattern of diffusion from box-office rich to box-office poor cinemas, and c) forecast the closing box-office of films that opened in the first-run cinemas from the opening Saturday night of the release.

The dataset that forms the basis of this proposal is drawn from the city of Philadelphia for the years 1935-36. Housed in the Warner Bros. Archive at the University of Southern California, I have recently uncovered the weekly billing sheets of 91 cinemas located in the city belonging to the Stanley Warner chain. The sheets provide micro data of an unparalleled nature about audience choices, consisting of daily box-office returns generated by the films screened at the 91 cinemas. This body of data has never previously been accessed. Once transcribed onto a database and analysed, it will make possible a much fuller understanding of audiences and the choices they made. Furthermore, the work will be directly applicable to contemporary consumer behaviour concerning experience goods.

I am looking for a graduate student that has an excellent quantitative historical background, who is excited about executing detailed fieldwork in the city of Philadelphia; building a dataset from archive materials using relational database software; and finally modelling the data to investigate consumer behaviour. Knowledge of a programming language is desirable.

ACEGES Software: video- World Oil Production

Below is video of World Oil Production for 5 Monte Carlo experiments. The first graphic is a stacked timeseries representing the contribution of each country - only countries that produce for the specific experiment are shown. The second graphic is shows oil production and reserve/production ration. I thank Maciej M. Latek for his help in the design of the graphics.

The first simulated year is 2002.


A better quality video can be seen at:
http://screencast.com/t/OWM4YzkzYz

ACEGES Software: video

Although this is not a good video (not sure how to convert swf to mpeg), it show the 'world oil production' and the production of 'Saudi Arabia' for certain Monte Carlo Experiments of the ACEGES.

The lines with the 'left colors' show the 'Production' and the lines with the 'right colors' shows the 'Remaining Reserves/Production'.

NOTE: the world oil production has not been modelled explicitly. This is the macro-regularity of the decisions made by the agents - the 93 countries modeled in ACEGES 1.0.



When the graphics show different combinations of colors, this signals a new simulation with different starting assumptions for key uncertain variables such as EUR, peak/decline point and oil demand.

The ACEGES Project: Presentation

This is a general presentation that I give about the ACEGES project.

ACEGES Software: Probabilistic Forecasts of World Oil Production to 2100

I have been asked by a few people to show a graphic of World Oil Production. This is given in the figures below. I have superimposed the 'peak year' based upon the simulation (Monte Carlo simulations based upon distributional assumptions of OIL EUR, Peak Point, Oil Demand).

Although the peak year and peak production contradict much of the existing views, a full explanation of the assumptions of the ACEGES program and the data used to initialize the simulations will be provided in the ACEGES 1.0 manual - expected by the end of July. The figures are based on only 134 simulations.



The blue dots is a random sample from the simulated data and the black lines (in the figure below) are estimated centiles using the Generalized Beta type 2 distribution. All the four parameters of the distribution were modelled using cubic smoothing splines. The actual production, red line, seems to follow the 2% centile value.

ACEGES Software: Smooth Centile Curves for United States

Below is the centile curves of the simulated oil production data for the United States. The distribution used for the centile curves is the Box-Cox t where all the four parameters (mu-location, sigma-scale, nu-skewness, tau-kurtosis) of the distribution were modelled (in GAMLSS) using a non-parametric Penalised Beta Spline.


Currently, we are fitting additional distributions to see if we can capture better the skewness and kurtosis of the simulated data as the above centiles are wobbly.

ACEGES Software: World Oil Production

The figures below show the simulated world oil production where simulation step 0 means year 2002. Note that the world oil production has not been modelled explicitly. The world oil production is the emergent pattern of the production behavior of the individual countries (agents). Even though these figures shows a few simulations (each line represents a single simulation), there is an interesting clustering of peak year and peak oil production.

ACEGES Software: Results for IRAN

Below are 4 figures of simulated data for Iran (black line is the actual production). The first (log representation) and third figures are based on Monte Carlo simulations of EUR, Peak Point and Demand while the second (log representation) and fourth figures are based on Monte Carlo simulations of EUR and Demand.

NOTE: These results are based on less than 100 simulations. Therefore, at this stage these figures should be treated as cartoons!

ACEGES software: Smoothing the Monte Carlo Oil Production data

Following from my last posting, these are some figures (Ecuador & US out of the 93 countries modeled) of the simulated data, generated using the ACEGES software and analysed using the R statistics software.

Black line is the actual production, blue dots is a sample of the 3.2 million (simulated) records (see last figure - US) and the density is based on the whole 'population' of the 3.2 million records.

ACEGES software: Preliminary Results

I have recently completed the ACEGES 1.0 software. This is an agent-based model of conventional oil production for 93 countries.

Due to the uncertainties with the EUR data and the oil demand, I implemented a Monte Carlo simulation that uses distributional assumptions of EUR data and oil demand based upon the data released from the Energy Information Administration, International Energy Agency, CIA and the US Geological Survey.

Next month, I will finish the User Guide and I will post the software on the CIBS website: http://www.londonmet.ac.uk/lmbs/research/cibs/cibs-scenario-planning/cibs-scenario-planning_home.cfm . If you want to access it earlier, please email me at v.voudouris@lodonmet.ac.uk

THe figure below sow the results from the US & Kuwait . RED line is the actual production while the black dots are the simulated data (77 simulation runs of 99 steps).

The ACEGES project: An ACE Model for the Availability of Global Conventional Oil Supply

I will present a paper at the 16th International Conference on computing in economics and finance. The abstract is given below:

The overall aim of this paper is to present a developing agent-based computational laboratory, termed the ACEGES (Agent-based Computational Economics of the Global Energy System) laboratory, for the systematic experimental study of the global energy system through the mechanism of Energy Scenarios. In particular, our intention is to show how Agent-based Computational Economics (ACE) and the Generalized Additive Models for Location Scale and Shape (GAMLSS) can be fused to help us understand better the challenging outlook for conventional oil supply by means of controlled computational experiments.

Specifically, the ACEGES laboratory, which is developed using the MASON (Multi-Agent Simulator Of Networks) library and the R software, models oil production curves to analyse the energy supply tensions at the national level. These production curves are semi-parametric regression models of i) original extractable oil, ii) remaining oil prior to the previous year’s production, iii) previous year cumulative production iv) previous year consumption and v) the net world demand left after an estimated demand increase, which can only be satisfied by Pre-peak Net Producer (PPNP) agent.


These semiparametric regression models are developed using the GAMLSS framework, which allows us to express macrovariables as statistical distributions. GAMLSS is a general framework of regression type of modelling in which the response variable can have a very general (up to four parameters) distribution and all of the parameters of the distributions can be modelled as linear or smooth functions of the explanatory variables. Thus, the R-based GAMLSS tool is used to build oil production curves, which are used as the decision rules of the agents. This also reflects that the R-based GAMLSS tool is integrated with the ACEGES laboratory as a way of providing a methodological advancement to undertake rigorous study of economic systems.

The Secret Life of Chaos at BBC iPlayer

Chaos theory has a bad name, conjuring up images of unpredictable weather, economic crashes and science gone wrong. But there is a fascinating and hidden side to Chaos, one that scientists are only now beginning to understand.

It turns out that chaos theory answers a question that mankind has asked for millennia - how did we get here?

In this documentary, Professor Jim Al-Khalili sets out to uncover one of the great mysteries of science - how does a universe that starts off as dust end up with intelligent life? How does order emerge from disorder?

It's a mindbending, counterintuitive and for many people a deeply troubling idea. But Professor Al-Khalili reveals the science behind much of beauty and structure in the natural world and discovers that far from it being magic or an act of God, it is in fact an intrinsic part of the laws of physics. Amazingly, it turns out that the mathematics of chaos can explain how and why the universe creates exquisite order and pattern.

And the best thing is that one doesn't need to be a scientist to understand it. The natural world is full of awe-inspiring examples of the way nature transforms simplicity into complexity. From trees to clouds to humans - after watching this film you'll never be able to look at the world in the same way again.