This project is primarily organized into a bunch of R scripts that clean data, analyze data, and then plot output. If you'd like to see how these files relate to one another, there is a nice visualization in file_dep_vis/file_dep_vis.svg.
These dependencies are also encoded in the Makefile, which is a utility that you should run the code from. There will be some example commands below that will allow you to execute high-level tasks.
Before any of the analysis code can be run, the C++ code must be compiled. Sometimes that can be tricky, particularly for users who have more experience using interpreted languages such as R.
To build the c++ code, you need:
- cmake to build c++ code
- Eigen3 for linear algebra functionality
- Boost for other math functionality
Catch2for the unit testspfa particle filtering libraryssmea more experimental particle filtering/MCMC library
The cpp code is in cpp/. After installing all the above software, you can build the project with CMake. Most IDEs handle CMake projects, and you can also do it from the command line.
Note: depending on where you installed the above software, you might have to chane a few lines in cpp/CMakeLists.txt. For example, say you downloaded eigen and didn't install it to /usr/include/eigen3. Then the line include_directories(/usr/include/eigen3) in cpp/CMakeLists.txt is wrong for you, and you need to change the path in there.
The cpp/ folder is a CMake project that can be edited with CLion. To open up the cpp/ code as a CLion project,
click here. From there, just hit the build button.
The (Debug) build command I use to build everything on Linux/Mac is
/path/to/cmake/bin/cmake --build /path/for/software/jsm22_pp/cpp/cmake-build-debug --target jsmpp_v2 -- -j 6
There is also a similar command to build the Release configuration.
If you build with CLion, you can run with CLion. This program requirements command line arguments, though. You need to specify the "run mode", and the path to the return observations. For help choosing the run mode, look at the comments in cpp/main.cpp. For help instructing CLion to use arguments when it runs the program, click here.
You can also run the program from your terminal. Here's an example of how to run the "Release" build. But remember, file paths may change, and you may change the "run mode."
./cpp/cmake-build-release/jsmpp_v2 1 "data/SPY_returns.csv"
You should run this command from the root directory because there are a few file paths hardcoded in the files :( R scripts that call c++ code follow this approach--they construct a command similar to the one above, and then pass it to system().
Once all the c++ code is built, you can start running everything my simply typing make in the command line. NB: you can also run R scripts on a case-by-case basis--to do that, follow the examples in Makefile (e.g. Rscript yadayadayada.R).