Réponse reçu.
Ca fait plaisir de recevoir une réponse plutôt complète, aussi rapidement.
Voici sa réponse (en VO), si quelqu'un veut traduire, libre à vous.
Dear Morgan,
>I glad to welcome you.
Thank you!
> First of all, your project is almost only based on stock price prediction asI can see on your website, which is then a finance based project.
In the past we've worked also in Social Network Analysis. Perhaps it will be again a relevant topic in the future.
But you are right, at the moment the focus is on Time Series Analyse. The financial application is just a nice to have. We are mainly interested in improving methods to discover patterns in changing environments. Usually data mining means looking for stable pattern in order to analyse current circumstances or for predicting the future. But these patterns ( or lets say rules or dependencies or influences) might change over time. When this happens common data mining methods get problems. We are looking for new methods that are able to deal with changing patterns.
> That's why a lot of members are worried about your project.
As you can see there is a scientific challenge behind the project. The stock forecasting application is used as an example for changing environments because patterns in stock market are changing. In addition, stock prices are public and free. Doing so, there are no costs to get relevant data. And last but not least: I am interested in stock market analysis.
> Can you explain me in a comprehensible manner what is the aim of your project and most of all what are expecting from that.
> Can you tell me a bit more about your project.
I hope I could give a brief description. I am going to write something more detailed on the website. But this takes a bit more time. For the moment I have a lot of work with stabilizing the project server and the distributed software. Sorry for the delay.
> You are using neural networks and genetic algorithms to compute and understand the market in order to build a model to outperform the market, right ?
Yes, Neural networks and other machine learning algorithms (Support Vector Machines, Decision trees ...)
> And, the most important thing, often, l'Alliance Francophone and a lot of others teams crunch on a project with only one condition; that your results will be published.
> Will you publish your result ? If you find something interesting, what will you do with it ?
As in the past, interesting scientific results will be published in articles, books or at conferences. I am also thinking about publishing stock forecasts. But that brings a lot of problems: e.g. legal problems and a huge implementation effort for the websites. But I like the idea to give some practical results back to the community.
Thank you for interest and your support
Nico Schlitter
Je trouve ce Nico sympathique. Et son projet semble être sérieux.
à tous
