将 sklearn 随机森林 model 转移到新服务器

Question

I built one model with sklearn RandomForestClassifier in an old server and now I need to migrate it to another server. How can I transfer the model to the new server? Which Python package should I use? Pickle? joblib? Thanks!

Answer 1

Use "joblib". Suppose your model is in a variable "my_model". Then the 'joblib' code would go like this:

# On your development machine
from joblib import dump
dump(my_model, 'model.joblib')

# On your new machine, following code would go to load the model
from joblib import load
my_model = load('model.joblib')

Note: Replace "model.joblib" with path to the model.joblib file.

Answer 2

pickle is the way to go

from sklearn import model_selection
from sklearn.linear_model import LogisticRegression
import pickle

# Fit the model on training set
model = LogisticRegression()
model.fit(X_train, Y_train) # fit on some data ...

# save the model to disk
filename = 'finalized_model.sav'
pickle.dump(model, open(filename, 'wb'))


# load the model from disk
loaded_model = pickle.load(open(filename, 'rb'))
result = loaded_model.score(X_test, Y_test) # predict some test data
print(result)