Calling python script from C#

Question

I have a C# code which helps to run python environment first and then it executes my python process. But the problem is it takes a lot of time to execute.

Actually i j

Answer 1

Why don't you actually use Python to run the code instead of embedding in C#? How are you going to deploy on another machine with Python dependencies?

If you would like to build machine learning models there are many frameworks like http://accord-framework.net/ for classic machine learning algorithms

Also try my project as well: deepakkumar1984/SiaNet (https://github.com/deepakkumar1984/SiaNet) Its a C# wrapper with CNTK backend. Trying to implement keras like wrapper. Hope it helps!

Answer 2

You'll need to create a new process in order to call your program.

look at this : C# equivalent to fork()/exec()

Answer 3

I would suggest you to use REST API to call python code from C# application. To achieve that you need to use two libraries: CPickle and flask

1. Expose line of code as a function and annotate
2. Serialise your model after training and load when predicting

Please refer to this code, I have created in python 3.5

from sklearn import datasets
from sklearn.ensemble import RandomForestClassifier
import pickle
from flask import Flask, abort, jsonify, request
import numpy as np
import json

app = Flask(__name__)

@app.route('/api/create', methods=['GET'])

def create_model():
    iris = datasets.load_iris()
    x = iris.data
    y = iris.target
    model = RandomForestClassifier(n_estimators=100, n_jobs=2)
    model.fit(x, y)
    pickle.dump(model, open("iris_model.pkl", "wb"))
    return "done"


def default(o):
    if isinstance(o, np.integer):
        return int(o)
    raise TypeError


@app.route('/api/predict', methods=['POST'])
def make_predict():
    my_rfm = pickle.load(open("iris_model.pkl", "rb"))
    data = request.get_json(force=True)
    predict_request = [data['sl'], data['sw'], data['pl'], data['pw']]
    predict_request = np.array(predict_request)
    output = my_rfm.predict(predict_request)[0]
    return json.dumps({'result': np.int32(output)}, default=default)


if __name__ == '__main__':
    app.run(port=8000, debug=True)

you can run it as: