save-load_Model.py

'''
Save and Load ML modules
A Script, to save model in specific file and load it later in order to make predictions. 
Serialize ML algorithms and save serialized format to a file.
Load file to deserialize models and use it to make new pridictions.
    - Pickle
    - Joblib
'''

from pandas import read_csv
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression

seed = 7

filename = 'pima-indians-diabetes.data.csv'
names = ['preg', 'plas', 'pres', 'skin', 'test', 'mass', 'pedi', 'age', 'class'] dataframe = read_csv(filename, names=names)
array = dataframe.values
X = array[:,0:8]
Y = array[:,8]
X_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size=0.33, random_state = seed)

# fit the model
model = LogisticRegressino()
model.fit(X_train, Y_train)



''' 1. Pickle '''
from pickle import dump
from pickle import load


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

# some time later...


# load the model from disk
loaded_model = load(open(filename, 'rb'))
result = loaded_model.score(X_test, Y_test)

print(result)




''' 2. Joblib '''
# Useful for some machine learning algorithms that require a lot of parameters or store entire dataset

from sklearn.externals.joblib import dump
from sklearn.externals.joblib import load

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

# some time later


# load the model from disk
loaded_model = load(filename)
result = loaded_model.score(X_test,Y_test)

print(result)