从 C++ 应用程序调用 python 多元线性回归代码

Question

I am doing my research in networking and I want to implement linear regression on my own data set(numeric). I used python for multivariate linear regression using sklearn library and now I want to embed/call that python implementation from my c++ code.

AS if I say that c++ is my application and python code is a service which provides services to my c++ code.... and I need pyhton code output in c++ code so I want to call python executable in c++ .

can I do that?

I search a lot as boost one is but I am not understanding how can I do... I used jupyter notebook for python and eclipse for c++.

x = final_tf[['LocalLoadHigh', 'LocalLoadLow', 'TransitLoadHigh',
    'TransitLoadLow','phaseTotalBlocking', 'phaseTotalLocalBlocking', 'phaseTotalTransitBlocking', 'PBlockingLocalHigh',
    'PBlockingLocalLow', 'PBlockingTransitHigh', 'PBlockingTransitLow',
    'UtilizationHigh', 'UtilizationLow']]
                # Create target variable, y

y = final_tf['WavelengthGroup']
 # Import model
 # Create linear regression object
lm = LinearRegression()
print(np.isfinite(x).all())
model = lm.fit(x,y)
predictions = lm.predict(x)
print(predictions)
lm.score(x,y)
lm.coef_

Answer 1

When you want different languages to collaborate, you have 2 possible ways.

1. the easy way: external program and communication through pipes/files:

   • you prepare the input data in your C++ program and save it to a file
   • you launch a python interpretor to execute your python script with fork/exec (on Linux), spawn (on Windows) or simply with system ; the python script reads the input file, apply the regression tools and save the results to an output file
   • you read the output file from your C++ application
   • alternatively, you could use pipes instead of plain files

2. the hard way: embed a Python interpretor in your C++ program and use the Python API or a tool like swift or sip to manage the passing of data from one language to the other.

   This will certainly be more efficient (less overhead of launching processes and serializing data) but the way to embed Python as a dynamic library depends on the actual OS. Refer to the Python documentation for your system.