Data Science

Galorath is more than a solutions provider to the Federal government. We are a data services and data science provider. As a leading provider of predictive analytics of cost and schedule for the Federal Government, Galorath is at the cutting-edge of the theory and practice of machine learning techniques. Our consultants’ toolset extends beyond Federal Government applications. Their expertise in collecting, normalizing, and analyzing data applies to a variety of domains, including healthcare, human resources, and customer data.

Galorath’s expertise in predictive analytics (“what will be”) also applies to prescriptive analytics (“what to do”). For example, Medicare penalizes hospitals for patients who are discharged and then re-admitted within 30 days. Predictive analytics can be used to predict what causes re-admittance. This information can be used in a prescriptive model to determine effective ways to reduce re-admittance.

Data Science: Domain Knowledge is an often overlooked but critical component

Machine Learning/Analytics

Machine learning/analytics is the combination of statistics and computer science

Data Science

Data science is where statistics, computer science and domain knowledge meet

MODEL DEVELOPMENT: Current Vs. Next Generation

Current state of cost models:

  • Separate models for cost and schedule – phasing done post-process/no parametric schedule models
  • Cost and schedule models are not integrated with engineering models
  • Apply methods designed for large data sets to small data set
  • Correlation-based models
  • Risk is weak/not rigorous
  • Earned value and cost are separate
  • Not robust – confuse signal with noise

Vision for Next Generation:

  • Integrated cost, schedule, and phasing models – parametric
  • Integrate cost within a model-based systems engineering framework
  • Continue development of methods for small data (Bayesian/imputation/component-level)
  • Causal modeling
  • Improve risk analysis (copulas/calibration)
  • Integrated cost and earned value/make earned value more rigorous
  • Incorporate cross-validation and other techniques to prevent overfitting

“Information is the oil of the 21st century, and analytics is the combustion engine.”