Home > cadastre > Neural networks, the best of Bolivia

Neural networks, the best of Bolivia

The return of Bolivia was tired, 22 hours and most complicated was to be at the last level stuck at Comalapa airport, El Salvador before coming to my boot’s country. It was a one fatigued week, hours of work from 8 to 5 seated almost all day, much food, but also much learning.

Almost everyone has concluded that the course has been too full of content and very little practical work; it affects the load on an instructor who must manage an all-day exhibition, with half boring PowerPoint’s and different level auditorium … half-asleep, the other half lost and a few looking for practical benefit to what we already do. However, the CD with presentations and exhibits complement of several countries has brought good results.

Among the papers, what most caught my attention is the application of neural networks to complex processes under the principle of artificial intelligence.

clip_image002

The problem

Whether it is done by a central institution or a local municipality to collect property taxes requires implementing a mass appraisal methodology. To do this there are several from simplified (lying) to simple (unsustainable). One of these methodologies is widely disseminated through the market method for valuing land and replacement cost for buildings. This requires at least three arduous tasks:

1. Update of improvements’ values. Its implementation is through what is known as building typologies, these are built with budget chapters, which in turn are integrated by construction elements and composed by basics by way of unit costs tokens. So that the simplest is to update the basis of inputs: materials, labor, equipment and machinery, plus professional services and then building typologies are ready to apply. Practicality of methodologies like this is that the field data collection for the valuation tab only requires calculating the construction area, design features, quality and conservation… all well-documented can overcome subjectivity.

For rural areas, it is also done a study of those features that give the building a productive value, such as permanent crops, marketable resources or potential use.

2. Map update of land values. This is built on a reliable sample of transactions in real estate, with significant representation and projected in time to have market value. Then these values ​​are converted into homogeneous areas containing a trend based on proximity and services.

3. Update public services networks. It happens that when changing road infrastructure state, to set an example, these characteristics affect a property in one or more of their foreheads. So it is ideal that values ​​move from the block to the street axis so that can be associated to the proportion that affects the building front… ideal, that area has certain characteristics that give it a value by service networks and close relationship to benefits which not only affect land value that can be very linear.

It is not difficult to do so every five years, but doing so in differentiated ways for many municipalities becomes unsustainable crazy even if there is a computer application, because it remains dependent on external data and field samples.

The application

Yedra García, from Spain economy Ministry, has presented a paper on the theme “Artificial intelligence applied to the mass valuation

The concept is there on the web, in English, however Yedra has raised a possibility, through the use of neural networks that applied to this problem would solve the automation methodology as complex as it may seem:

It means that a minimum amount of middle-level indicators, may have a comparative relationship that when send down an inputs trend of values and up a values’ tentative proposal of homogeneous areas through spatial analysis by conditions’ similarity, can generate a matrix that makes redundancy on both tracks against real data, such as construction prices newsletters or real estate values.

Of course this does not carry a simple tabular data analysis, but also a layers spatial analysis which affect recovery, trunk roads interconnection and topological analysis of shared neighborhood.

This could bring results beyond mere valuation for property tax purposes, like the construction or planning based on impact conditions in revaluation and value capture … among others.

clip_image004

The position let me envy of green smoking someday on the intention to implement it

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.