Preliminary Detailed Design
Table of Contents
Team Vision for Preliminary Detailed Design Phase
- Understand most of the current code, research
literature and focus of the project.
- Have a pseudo code of what we want/need to have.
- Meet with subject matter expert about suggestions and
Design and Flowcharts
1. Segment the countries into many markets (ideally: each
country is a market).
2. Create a 2^k Factorial Experiment, where k is the number of markets and 2 represents whether a market still has a budget or not.
3. Using the UNICEF data, extract a data set that reflects the budget uncertainty of a market.
- This may require building a predictive model to
estimate a market's budget uncertainty.
4. Plot the data on a Cullen and Frey Graph to determine
what theoretical continuous statistical distributions the
data may come from.
6. Plot the fit and the data set together to graphically validate goodness of fit for each promising distribution.
- Kolmogorov-Smirnov statistic
- Cramer-von Mises statistic
- Anderson-Darling statistic
- Aikake's Information Criterion
- Bayesian Information Criterion
8. If a promising distribution has enough evidence to
suggest that the data set comes from that distribution,
then begin simulating ABP results by randomly sampling
from the promising distribution.
9. If a promising distribution fails to have enough evidence, then fit an empirical distribution on the data set and then begin simulating ABP results by randomly sampling from the empirical distribution.
Plans for next phase
- Execute prototype design.
- Evaluate ABP results.
- Develop a Test Plan to test how well Engineering
Requirements are met.
- Determine how to improve the prototype design.