Victor-SQL integrates incremental schemes with an RDBMS via a (hopefully) easy-to-use python interface.

**How to Get Started:**

Victor is very easy to setup and use for solving a variety of statistical problems. In the documentation tab we explain the Victor syntax. You can also find the Victor installation guide , view the Python documentation, and unit tests coverage report. In the examples tab you find examples of using Victor for solving a variety of statistical problems. Finally, you can download the Victor source code, installation guide, and examples from the download link.

**What Victor Tries to Solve:**

Database researchers and the database industry have recently begun to incorporate statistical techniques into relational database management systems (RDBMS). Incremental projected gradient methods (iPGMs) are a broadly applied numerical optimization technique for solving large-scale data-intensive problems, but they have not yet been integrated into an RDBMS. The Victor project demonstrates that iPGMs can be implemented with data access properties similar to standard SQL queries.

**How Victor Works:**

Victor is a middle-ware system that uses an RDBMS to solve a large class of data analysis problems. Victor issues statements to PostgreSQL and registers a collection of user-defined functions with the RDBMS that perform the iPGM computation. A user interacts with Victor by defining a set of views. Victor translates these views to standard tables in PostgreSQL. Since the tables are standard PostgreSQL tables, the user may use them as standard SQL queries in their application.

**What Victor Can Achieve:**

Since Victor makes use of RDBMS, it can leverage 30 years of RDBMS optimization technology to efficiently compute iPGMs inside an RDBMS. Through experiments on several real-world and benchmark data sets, we have demonstrated that Victorâ€™s run-time performance is comparable to specialized solutions for many classification and regression problems. Moreover, for some data analysis problems Victor is orders of magnitude more efficient.