Phillip Colla, Algorithmist, Data scientist, Developer

About

I have 38 years experience implementing and commercializing predictive analytic engines, such as CART, TreeNet (i.e., Stochastic gradient boosting aka "MART"), RandomForests, MARS, GPS (Generalized PathSeeker), LOGIT (binary and multinomial logistic regression), CTM (Continuous Time Modeling), and Survival Analysis among others. I have considerable mastery of the algorithmic and computational details of these engines having taken all of them from their earliest academic origins (e.g., CART in 1988, originally written in Mortran/Fortran by the CalStat team at Stanford and Cal Berkeley) through many cycles of extension, testing and refactoring into successful commercial products, most notably Salford Systems' "Salford Predictive Miner" and, following Minitab LLC's acquisition of Salford Systems, into Minitab Statistic Software. Over the years I have rewritten parts or all of these engines, sometimes multiple times, as market and client requirements have evolved while remaining faithful to the original algorithm and spirit of the authors.

CART, TreeNet/MART, RandomForests and MARS are some of the most important and foundational algorithms, and their authors some of the most brilliant minds, in the data mining and predictive analytics space. I acquired mastery of all these methods as both an algorithmist and developer in order to bring them to market, transforming them from early academic implementations into stable, fast, well-tested, computationally-correct and commercially-successful software products. Our efforts led to Salford Systems' flagship software "Salford Predictive Miner" being licensed to over half of the Fortune 500 and being cited often in scholarly literature. Since 2017 I have been a lead developer responsible for deploying all-new implementations of Salford Systems' predictive analytics engines into Minitab Statistical Software, and we have had great success. I work primarily in C++ but have decades of experience with Fortran and have been a part of teams using Perl, Python, PHP, Mysql and Javascript among other languages as well.
Selected citations of software publications with Dan Steinberg / Salford Systems
  • CART: Classification and Regression Trees, a tree-structured non-parametric data analysis
  • Logistic Regression, Systat (2004)
  • MARS user guide, Salford Systems (1999)
  • SURVIVAL: A supplementary module for SYSTAT (1988)
  • LOGIT: A supplementary module for SYSTAT (1991)
Advisory Algorithm Engineer
Minitab, LLC, State College, PA
March 2017 - present (5 years)
Senior / Lead Developer
Salford Systems
1984 - 2017 (33 years)
Contact
My name as one word, two L's in first and last names, at Gmail
+1 760 707 7153
Master of Science, Engineering Science, 1986
University of California, San Diego
Summa cum laude
Bachelor of Science, Mechanical Engineering, 1985
University of California, San Diego
Summa cum laude, 1st in graduating class