wine_model.py

import pandas as pd
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import cross_val_score
from sklearn.model_selection import cross_val_predict
from sklearn import metrics
import coremltools


dataset_url = 'http://archive.ics.uci.edu/ml/machine-learning-databases/wine/wine.data'
names = ['cultivar', 'alcohol', 'malic_acid', 'ash', 'alkalinity_ash', 'magnesium', 'total_phenols', 'flavonoids', 'nonflavonoid_phenols', 'proanthocyanins', 'color intensity', 'hue', 'od280_od315', 'proline']
data = pd.read_csv(dataset_url, names=names, header=None)


X = data[['alcohol','malic_acid', 'ash', 'alkalinity_ash', 'magnesium', 'total_phenols']]
y = data['cultivar'].astype(str)


# Create the model
model = RandomForestClassifier()

# Evaluate the model with cross validation
scores = cross_val_score(model, X, y, cv=5)
print('Scores: {}').format(scores)
print('Accuracy: {0:0.2f} (+/- {1:0.2f})').format(scores.mean(), scores.std() * 2)

predicted = cross_val_predict(model, X, y, cv=5)
print('Predicted: {}').format(predicted)
accuracy_score = metrics.accuracy_score(y, predicted)
print('Accuracy: {0:0.2f}').format(accuracy_score)


# Fit the data
model.fit(X, y)


# Convert model to Core ML 
coreml_model = coremltools.converters.sklearn.convert(model, input_features=['alcohol','malicAcid', 'ash', 'alkalinityAsh', 'magnesium', 'totalPhenols'])

# Save Core ML Model
coreml_model.save('wine.mlmodel')

print('Core ML Model saved')