Description:
My project is to visualize the hidden Markov chain structure of data set from educational study. The variables of the model are discrete, and the parameters are estimated via Bayesian model.
Data was provided by my friend working in Educational Test Service, along with some old data sets. They provide the statistical model so what I need is to use their models instead of analyzing data by myself.
Goal:
The result is for statisticians in educational field, so I hope my work will be able to understand by people in this domain. The visualization will describe the HM chain, parameters, important transit states and probabilities.
Initial reading:
1. JIAN ZHANG, MARKOV CHAIN AND HIDDEN MARKOV MODEL.
This reading is to understand the basic idea of fundamental HMM.
2. Gerard Rambally, A Hybrid Visualization Hidden Markov Model Approach to Identifying CG-Islands in DNA Sequences.
This reading is to learn how to explain HMM to people in other domain, like biology and genetics.
3. Valeria De Fonzo, Filippo Aluffi-Pentini and Valerio Paris, Hidden Markov Models in Bioinformatics, Current Bioinformatics, 2007, 2, 49-61
This reading provides a good example to visualize HMM with emphasizing probabilities in the model.
4. Some coding skill. I think it is important for me.
Schedule:
Week 1 (04/16): Try to implement the reference related to the data and model. Read the papers to understand HMM.
Week 2 (04/23): Provide a visualization of the data, which at least has the basic features of HMM. If possible, I would get some feedback about what to do to improve it.
Week 3 (04/30): I hope to get some more visualization derived from the initial one and compare them.
Week 4 (05/06): Finish the implementation and make a function/package/toolbox that could be used for later analysis. Write a report.