[Analyzer, SubGroup 2] S.M.A.R.T. Goals

[lauraccunningham]

Members:
Laura CUNNINGHAM (Operational Lead) @lauraccunningham
Disi KOA (Analyzer) @gnolnait
John RISKO (Analyzer) @johnrisko
Tay SHIN (Technical Lead) @taywon

S.M.A.R.T. Goals

• [S] Specific: As a group, John RISKO, Tay SHIN, Disi KOA, and I, want to produce a working function that incorporates the R packages ETAS, SAPP, and PtProcess, to understand a way of measuring error rates of alarm coverage by comparing our outputs with the MDA and simple Stark models. From the curators, we need a set of data to defining the temporal and magnitude aspects of our model, so we can compare, analyze, and help the visualizers produce visual representation of error rates of alarm coverage. To develop a better understanding of earthquakes models, we need to find a optimal relationship between time and error so we can understand earthquake modeling more accurately.
• [M] Measurable: Our results will conclude with a optimal result of which model best predicts the occurrence of an earthquake. Although our course has a definite end date, mid-December, our ability to understand the various models will adapt and change with time. We will have to halt our studies with a prediction of which model best suits our needs, yet can continue to be compared with changing data as time goes on.
• [A] Attainable: With coordination between the Curators & Visualizers we hope to use the .ppx file produced to understand parameters and outputs as a representation of earthquake model alarm error. Using R packages we want to accomplish a developed model of interesting information developing a clear relationship between time and error of earthquake model alarms.
• [R] Relevant: As of now, we can produce sudo code and a simple subset to prepare a larger function we hope to run our model through in the future with the particular .ppx file given to us by the Curators. By looking at the various model types and understand the parameters and outputs clearly is where we can begin to understand how we are going to tackle out project. In the future we will use our acquired knowledge to understand where we will be getting out inputs from the .ppx and manipulating through analysis for our ETAS evaluation.
• [T] Time-Bound: The semester concludes 18 December 2013.

Determining the domain and R packages to be used will be crucial in the formation of data we hope to evaluate.

Weekly Goals

• Use the SAPP program to estimate the model parameters:
• Week 1:
  • Simulate a model using the ETAS model (Optional Step)
    • This test the strength of our model against a data set created from the ETAS model.
• Week 2:
  • Estimate conditional intensity over time. This step will help us create our alarms.
  • Set up alarm function for ETAS from level sets of the conditional intensity over time.
• Week 3:
  • Calculate the error rate through the simulated or real data set.
  • We send to a visualizer group our error rate.