Pipeline deployment in Flask (python) - machine-learning

Question

I'm trying to deploy my model built using Pipeline via Flask, however I'm facing the following Attribute error
'Can't get attribute 'FeatureSelector' on main' from 'app.py''
Here is my code for model.py:
(After loading the necessary libraries and reading the data, I have defined the class for my pipeline)
class FeatureSelector( BaseEstimator, TransformerMixin ):
#Class Constructor
def __init__( self, feature_names ):
self._feature_names = feature_names
#Return self nothing else to do here
def fit( self, X, y =None):
return self
#Method that describes what we need this transformer to do
def transform( self, X, y = None):
return X[ self._feature_names ]
LE = LabelEncoder()
class CategoricalTransformer( BaseEstimator, TransformerMixin ):
#Class constructor method that takes in a list of values as its argument
def __init__(self, cat_cols = ['Response', 'EmploymentStatus', 'Number of Open Complaints',
'Number of Policies', 'Policy Type', 'Renew Offer Type',
'Vehicle Class']):
self._cat_cols = cat_cols
#Return self nothing else to do here
def fit( self, X, y = None ):
return self
#Transformer method we wrote for this transformer
def transform(self, X , y = None ):
if self._cat_cols:
for i in X[cat_cols]:
X[i]= LE.fit_transform(X[i])
return X.values
class NumericalTransformer(BaseEstimator, TransformerMixin):
#Class Constructor
def __init__( self, MPA_log = True):
self._MPA_log = MPA_log
#Return self, nothing else to do here
def fit( self, X, y = None):
return self
#Custom transform method we wrote that creates aformentioned features and drops redundant ones
def transform(self, X, y = None):
if self._MPA_log:
X.loc[:,'MPA_log'] = np.log(X['Monthly Premium Auto'])
X.drop(['Monthly Premium Auto'], axis =1)
return X.values
I have created different pipelines for numerical and categorical festures. They have been combined using Feture Union in Full Pipeline.
full_pipeline = FeatureUnion( transformer_list = [ ( 'categorical_pipeline', categorical_pipeline ), ( 'numerical_pipeline', numerical_pipeline ) ] )
X = df.drop('Customer Lifetime Value', axis = 1)
y = df['Customer Lifetime Value']
y = np.log(y) #Transforming the y variable
full_pipeline_RF = Pipeline( steps = [('full_pipeline', full_pipeline),('model',
RandomForestRegressor(max_depth=21, min_samples_leaf= 8, random_state=0))])
full_pipeline_RF.fit(X, y)
# Saving model to disk
pickle.dump(full_pipeline_RF, open('model.pkl','wb'))
# Loading model to compare the results
model = pickle.load(open('model.pkl','rb'))
Model has been called in app.py file with the following code:
import numpy as np
from flask import Flask, request, jsonify, render_template
import pickle
app = Flask(__name__)
model = pickle.load(open('model.pkl', 'rb'))
#app.route('/')
def home():
return render_template('index.html')
#app.route('/predict',methods=['POST'])
def predict():
'''
For rendering results on HTML GUI
'''
int_features = [float(x) for x in request.form.values()]
final_features = [np.array(int_features)]
prediction = model.predict(final_features)
output = round(np.exp(prediction[0]),2)
return render_template('index.html', prediction_text='Customer Lifetime Value $ {}'.format(output))
if __name__ == "__main__":
app.run(debug=True)
The code works fine in Jupyter. Even while running in Spyder, it doesnt throw any errors. Please help me with this code, I'm stuck only on the execution bit.

Answer 1

This was actually simple. All i had to do was jn my app.py I had to pass the classes created during pipeline.
These are new classes customised for the case, hence those classes need to be pass.
Simply write 'pass' after every class is defined.