Carlos's Escience Blog Big Data Econometrics

Weeks 4 and 5

Oct. 21,23,28,30

  • Developed STATA code to extract full sample from airline dataset
  • Tested full sample on EC2 instance r3.2xlarge (memory-optimzed, 61GB RAM)
  • Simulation consumed full 61GB RAM and failed
  • Andrew diagnosed that my input matrices were dense, and that I could save RAM by making them sparse
  • Searched for R packages that would run desired statistical estimation techniques using sparse matrices as inputs
  • Found diverse packages capable of computing desired estimation techniques using sparse matrices as inputs
  • (In process) Revising R code to utilize these packages