- The ordinary least squares method introduces a bias that can be very high if the cost data are scattered, to the point that this method was eventually called ” linear regression “: concretely the found relation “regresses” towards the average of the costs. Practically it implies that the found relation underestimates the “heavy” products (greater than the average, the average being defined here as the gravity centre of the data) and overestimates the “light” products (lower than the average)

- It also tends to neglect the light products, in the sense that it favors the “heavy” products to the detriment of the others. This results from the fact that, concerning costs, the data precision of costs is never given in value but in percent; strong costs therefore have a less good absolute precision. Now the method tries to minimize the squares of the distances: it thus is more naturally interested in the high costs.

- Finally the « outliers» can seriously modify the found relation. The outliers are points for which the cost is generally (but not always) very different from the others: they present, compared to the others, important distances that the method tries to reduce, and therefore “pulls” the relation’s curve towards these points..

Regarding costs – unless the costs are well aligned (which doesn’t happen very often) – the ordinary least squares method has serious inconveniences, that many authors tried to minimize by bringing mathematical corrections (these exist in EstimLab for persons who really want to use this method). Our idea was, rather than to look for palliatives, to replace it by a method that the computer facilitates nowadays. This method – generalized in the median space – consists in not minimizing the squares of the distances, but the absolute distances between the data and the relation we are looking for. This new method is only interested in the position of the costs in regard to the relation instead of the cost values . With this method, the inconveniences mentioned above disappear automatically and the palliatives become totally useless!

And the found relation is better than the one produced by the OLS method, in the sense that it is closer to data …

This method leads to mathematical procedures (used in EstimLab) completely different to the previous ones. For example: for the OLS method, Gauss showed that the deviations (between the cost data and the relation) must be distributed according to the curve known under the name of “bell curve” (or in more mathematical terms of ” curve of Gauss »), while the new method has to see the deviations distributed according to a curve with a ” witch’s hat ” form (or Laplace » curve ). Consequently the deviations, on average, are lower. This also leads to a different calculation of the confidence intervals for the estimated costs (and all this is in EstimLab).

EstimLab thus brings a better solution regarding the search for the relation between costs and significant variables, and therefore estimated values of better quality for new products.

]]>

Our softwares can help you to deal with your data correctly.

If you have little data, MAP-H is the software, at least for mechanical, et/ou electric, and/or electronic products, whatever their functions are, whatever size or application domain (from a consumer product to a product for space as well as nuclear or aircrafts …). MAP-H allows you to estimate, with buil-in algorithms (these can be adjusted to your needs) all the life cycle costs : development, production, maintenance, modifications.

If you have more data, classify them by familiesand add the relatif characteristcis for each product (or software, or activity, …). EstimLab will help you estimate very quickly the cost of a new item belonging to that family.

La manière de réaliser cette estimation est – pour chaque famille – de développer une A mathematical relation between the cost and the products’ characteristics, is developed for each family in order to to make an estimation possible. Many procedures are available to first analyse the data quickly so any abnormality that could disrupt the quality of the relationship. Once the abnormalities found – and resolved – the search for a relation can start. The starting point is, if each product is defined by several characteristics, to find out which charcateristics really determine the cost (EstimLab has several procedures to help you and they obly take a few seconds). You then choose the kind of relation you wish (amongst four) unless you can’t really choose, which then leads you to another procedure. If that choice is possible, EstimLab will calculate the relation by trying to make it as close as possible to your data ; this takes less time than one minute! Two procedures are available to get there

- the classical one, rather obsolete now – it existed in the 19th century – consists in finding the relation’s coefficients by minimizing the sum of squares of residuals between this relation and the data. This procedure has – at least in cost domain – several serious disadvantages ; some authors have tried to make up for these disadvantages, but these are only stopgaps …

- the modern one – entirely developed in our tool – consists in minimizing the sum of the absolute values of the distances. It is an extension of the « median » very well used by the statisticians when analyzing distributions.

Once the relation found, cost estimation of any new product, for which you only need to know the characteristics that are « cost drivers », can be done in one second. EstimLab also give you immediately the confidence interval (customizable) of the estimated cost, taking in account, if necessary, the uncertainty – that can still exist when the estimation is required – regarding the values of the characteristics.

]]>These variables can be many as the software accepts up to 30 quantitative variables (to which one can associate a number) and 16 « qualitative » variables, each one having several « modalities ».

These quantitative variables are very important in cost estimating and can rarely be substituted.

Because of their nature, one can’t associate a number to those variables. Each one has “modalities”. For example for a supplier, the modalities will be the name of each supplier; another qualitative variable could be the product’s material, in which case the modalities could be a list such as steel, aluminium, titanium etc …

*How are these variables used in conjunction with the quantitative variables * (such as mass, power or flow). Two possible solutions

- the first one consists in considering that the relation between the cost and the quantitative variables remains the same for each modality in the set of the qualitative variables. The influence of these qualitative variables just shifts, more or less, the relation between the cost and the quantitative variables. This is the easiest solution.

- the second implies that each modality (for the given set of qualitative variables) can disturb the relation. One then searches for as many relations as existing modalities.

Now how can one decide between these two alternatives ?

EstimLab has a procedure that gives you the parts of the answer with just one click.

]]>Volume 1: The Parametric Approach

**Volume 1**: “The Parametric Approach” cost estimating is introduced; the preparation of data before utilization is discussed; the basic concepts of ‘general’ cost estimating models are examined; the use of cost models is considered and finally, risk analysis is introduced.

Contents: Part 1: Cost Estimating: A definition.- Cost Measurement.- Overview of Cost Estimating.- ‘Elementary Cost Estimating’.- An Introduction To Parametrics for the Beginner.- Part 2: Data Collection.- Economic Conditions and Currencies.- The Cost Improvement Curves.- Plant Capacity and Workload.- Other Normalizations.- Part 3: Definition of a General Model.- Building a General Model.- New Concepts in General Modelization.- Modelization in Other Classes.- A Word about the Future.- Part 4: Using a Specific Model.- Using a General Model.- Introduction to Risk Analysis.- Auditing Parameters.

2006 XVIII, 299 p. 88 illus. Hardcover

Decision Engineering

ISBN 1-85233-973-X

**Volume 2**: “Building a Specific Model” focuses on the building of ‘specific’ cost estimating models. Understanding the procedures is key in creating successful specific models; so all these procedures (classical and new) are described in this volume.

Contents: Part 1: Population and Sample.- Describing a Population.- Typical Distributions.- Part 2: Dealing with One Variable Only.- Data Analysis of Two Variables.- Simultaneous Data Analysis on J Quantitative Variables.- Working with Qualitative Variables.- Part 3: Finding the Center of the Cost Distribution : Choosing a Metric.- Looking for the Dynamic Center: The Bi-linear Cases.- Using Several Quantitative Parameters: The Linear Cases.- Using Qualitative Variables.- Non-linear Relationships.- Part 4: Studying the Additive Residuals.- The Other Residuals.- Part 5: From Sample to Population.- Building the Model.

2006 XXXVI, 362 p. 171 illus. Hardcover

Decision Engineering

ISBN 1-84628-042-7

But

- how do you deal with the qualitatives ?
- how do you normalize your data that don’t have the same currency ?

EstimLab handles those questions ….

]]>