What we can do

We can build all sorts of data mining techniques into your websites and applications. We can also build standalone classifiers for you and/or apply them to your datasets. For that matter, we can do software and web development for you without doing any mining (and at very competitive rates). Here are some of the mining techniques we can use:

• Classification
  • Binary
  • Multi-class
  • One-class (Data Description, Outlier detection)
• Clustering
  • Partition-based
  • Grid-based
  • Hierarchical
  • Model-based
  • Spectral
• Regression
• Association Rule Mining
• Density Estimation
  • Parametric
  • Nonparametric
• Feature Extraction
  • Linear
  • Nonlinear
  • “Manifold Learning”

And the languages we can work with:

• C++
• Java
• Perl
• C#
• Visual Basic
• Matlab
• Prolog
• XHTML
• CSS
• Javascript
• Actionscript
• SQL

We can give you a guaranteed minimum accuracy upon a free consultation, provided that you give us access to the dataset in question. We cannot, however, guarantee higher accuracies than the one we quote at the beginning, nor can we attempt to code to attain a specific accuracy level specified by the client. We can tune most classifiers to trade off between true positive and false positive rates if you have a specific balance in mind, but the granularity of the adjustment may be out of our control.

We can code your entire application from scratch in addition to enhancing it with mining. Of course, this will take a bit more time and will cost a bit more, depending on the size of the application.

We can work with networked applications and agent-based systems.

We can develop on most common platforms, including Windows, Linux, Mac OS X, and various Unix-based systems.

We can (and will) write very clean and optimized code.

We can support our projects. Maintenance is free for one year from the completion date on our projects. If the project is over a year old, has been created or modified by someone other than us, or requires new features rather than fixes, we'll still provide support for a fee.