MH370_MCMC

Modeling the Last Flight of MH370 with a Markov Chain Monte Carlo Method

---

Versions

Version 3:

http://nbviewer.ipython.org/github/myhrvold/MH370_MCMC/blob/master/MH370_MC_ConorMyhrvold-V3.ipynb?create=1

---

Version 2

Part 1 of 3

http://nbviewer.ipython.org/github/myhrvold/MH370_MCMC/blob/master/MH370_MC_ConorMyhrvold-V2-Part1.ipynb?create=1

Part 2 of 3

http://nbviewer.ipython.org/github/myhrvold/MH370_MCMC/blob/master/MH370_MC_ConorMyhrvold-V2-Part2.ipynb?create=1

Part 3 of 3

http://nbviewer.ipython.org/github/myhrvold/MH370_MCMC/blob/master/MH370_MC_ConorMyhrvold-V2-Part3.ipynb?create=1

Also see repo for full version (downloadable, ~17 MB, and ~250KB w/ no output)

---

Version 1 fixed:

http://nbviewer.ipython.org/github/myhrvold/MH370_MCMC/blob/master/MH370_MC_ConorMyhrvold-bug_fixes.ipynb?create=1

---

Version 1 original:

http://nbviewer.ipython.org/github/myhrvold/MH370_MCMC/blob/master/MH370_MC_ConorMyhrvold.ipynb?create=1

---

Articles

Accompanying Fast Company Labs stories:

http://www.fastcolabs.com/3028265/how-i-narrowed-down-the-location-of-malaysia-air-using-monte-carlo-data-models

http://www.fastcolabs.com/3028331/more-about-our-methodology-tracking-mh370-with-monte-carlo-data-models

http://www.fastcolabs.com/3028569/this-data-model-shows-mh370-could-not-have-flown-accidentally-to-its-destination

http://www.fastcolabs.com/3028847/with-new-information-our-data-models-point-to-foul-play-on-malaysia-air-flight-370

http://www.fastcolabs.com/3029199/as-black-box-pings-go-silent-heres-how-data-can-narrow-the-search-for-malaysia-air-flight-37

---

Update Log

Updates 3/27:

For a standalone Python script version, see the work of Nathaniel Laughlin (github username natlaughlin): https://github.com/natlaughlin/MH370_MCMC (thanks! :0) )

---

Updates 3/29:

Working on Version 2 -- in course of doing that, realized and corrected 3 bugs in Version 1.

Updated Version 1 while keeping original, for comparison purposes and transparency.

---

Updates 3/31:

Added Version 2. Visualize via nbviewer in three parts (file size too large otherwise). Part 1, Part 2, Part 3.

Download the 17 MB file with V2 at the end, or with no output (named as such). Those are the ones you'll want to re-run the code yourself.

---

Updates 4/6:

I'm working on more analysis, and formally trying more parameter variations from my current model. I aim to have that done by mid-April. After that, unclear that I'll be able to get greater resolution of search areas besides what I've currently done. I'll further clean up the code, in a separate notebook, with the next set of analysis.

---

Updates 4/8:

Article summarizing what I've done to date, in addition to context of latest MH370 news, added to FastCo Labs coverage. "With New Information, Our Data Models Point To Foul Play On Malaysia Air Flight 370".

---

Updates 4/14:

Version 3 finished and released. Where to continue the search, based on Version 1 and Version 2 implementations and assumptions.

---

Updates 4/15:

Small typos corrected in text.

---

Other

Feel free to reach out with comments, critiques, criticisms.

To view IPython notebook in browser copy .ipynb url in here: http://nbviewer.ipython.org/