CODE
Econometrics
Fedora Core is my favorite Linux distribution
R for simple data manipulation, summary statistics, simple regressions
Computational econometrics packages:linear regression models, microeconometrics, nonparametric methods, linear models for panel data, GMM, time series, etc.
Java for everything else where speed is important: estimation of demand, dynamic games
java.sun.com for Java foundation (compiler, libraries, executor)
Parallel Colt is a multithreaded version of Colt - a library for high performance scientific computing in Java. It contains efficient algorithms for data analysis, linear algebra, multi-dimensional arrays, Fourier transforms, and statistics.
JAMA for matrix library
Ports of Fortran optimization libraries (UNCMIN, MINPACK) at FPL Statistics Group
IPOPT -optimizer for large scale optimization problems (including constraint optimization problems)
The Apache Commons Mathematics Library - It contains classes for numerical analysis, statistics, optimization, graphics, etc.
MySQL for advanced data manipulation and large databases. It works very well with R and Java.
Stata for simple regressions when I do not have R
Why R?
Cost: Free!
Superior graphics
Object orientation
Reproducibility
Excellent programming support
Why Java?
Cost: Free!
Excellent documentation
Object-oriented
Automatic memory allocation
Dynamic optimization of execution paths
Designed from day one to be networked, multithreaded, write-once-run-anywhere cross-platform
Update January, 2018 (extra job)
I am working on a Java package (eio) for structural econometrics based Parallel Colt and The Apache Commons Mathematics Library. Features:
Easy work with panel data in Java (summary statistics, data selection, merging data )
OLS, 2SLS, GMM estimation
Logit, multinomial logit estimators
B-Splines
Chebyshev interpolation (using vectors and matrices of points)
Classes for approximate dynamic programming (large state space)
Function interpolation
Demand estimation (Nested Logit)
Productivity estimation
Different classes that are useful when estimating dynamic games
Jacobian Matrix
Numerical Solvers (Broyden, Newton, Over-identified large system of equations)
Copyright (C) Florin Maican, 2024-2025