CODEWITHSUNDEEP
pythonmachine-learningscipylinear-regressionmlflow
RJ-mac
RJ-mac

Reputation: 33

How can I apply mlflow to handle scipy models?

I have implemented a ML model using scipy in Python. This model solves a linear regression estimation problem that constraints the regression weights to be inside a given interval.

Once the model is calibrated, I store the weights returned by scipy.optimize and use them like this in order to predict new samples:

import numpy as np
def predict(scipy_model, x_test):
    w = scipy_model.x
    y_pred = np.sum(w * x_test, axis=1)
    return y_pred

I want to deploy this model in a production environment using mlflow. However, I haven't been able to see how to integrate scipy with mlflow in the docs.

If it is not possible, can I create a custom "Model" class that has custom train and predict functions and integrate it with mlflow?

Upvotes: 0

Views: 39

Answers (0)

Related Questions