Getting KeyError in Klepto when used from local machine

Question

I want to persist my ML model to my local machine. I have followed https://stackoverflow.com/a/29291153/11543494 this answer to store ML model in local machine but when I load persisted ML model from local machine I get Key Error.

I have created ML model which will predict category of URL. Now I want to integrate ML model with my web application and that is why I created API using Flask.

I have tested my ML model in jupyter notebook where I have my code related to ML model now, I just want to dump my ML model and use it in my API. In jupyter notebook I am getting proper output but when I load persisted file in my API code, I get KeyError. I tried using pickle, joblib but there I am getting MemoryError, I tried to resolve that also but I was unable to solve that issue so I am trying Klepto.

Klepto code

from klepto.archives import dir_archive

model = dir_archive('E:/Mayur/Sem 5/Python project/model_klepto',{'result':gs_clf},serialized=True, cached=False)
#gs_clf = gs_clf.fit(x_train, y_train) #RandomizedSearchCV
model.dump()

API code

import numpy as np
from flask import Flask, request, jsonify, render_template
from klepto.archives import dir_archive

app = Flask(__name__)

demo = dir_archive(
    'E:/Mayur/Sem 5/Python project/model_klepto', {}, serialized=True, cached=False)
demo.load()


@app.route('/')
def home():
    return render_template('index.html')


@app.route('/predict', methods=['POST'])
def predict():
    input = request.form.values()
    final_feature = [np.array(input)]
    prediction = demo['result'].predict([str(final_feature)])

    return render_template('index.html', prediction_text=prediction)


if __name__ == "__main__":
    app.run(debug=True)

When I run API I get KeyError:'result'.

If I run below code in jupyter notebook, I get correct output

demo = dir_archive(
    'E:/Mayur/Sem 5/Python project/model_klepto', {}, serialized=True, cached=False)
demo.load()

demo

Output>

ir_archive('model_klepto', {'result': RandomizedSearchCV(cv='warn', error_score='raise-deprecating',
                   estimator=Pipeline(memory=None,
                                      steps=[('vect',
                                              CountVectorizer(analyzer='word',
                                                              binary=False,
                                                              decode_error='strict',
                                                              dtype=<class 'numpy.int64'>,
                                                              encoding='utf-8',
                                                              input='content',
                                                              lowercase=True,
                                                              max_df=1.0,
                                                              max_features=None,
                                                              min_df=1,
                                                              ngram_range=(1,
                                                                           1),
                                                              preprocessor=None,
                                                              stop_words=None,
                                                              strip_accen...
                                                               sublinear_tf=False,
                                                               use_idf=True)),
                                             ('clf',
                                              MultinomialNB(alpha=1.0,
                                                            class_prior=None,
                                                            fit_prior=True))],
                                      verbose=False),
                   iid='warn', n_iter=5, n_jobs=None,
                   param_distributions={'clf__alpha': (0.01, 0.001),
                                        'tfidf__use_idf': (True, False),
                                        'vect__ngram_range': [(1, 1), (1, 2)]},
                   pre_dispatch='2*n_jobs', random_state=None, refit=True,
                   return_train_score=False, scoring=None, verbose=0)}, cached=False)

demo['result'].predict(['http://www.windows.com'])

Output> array(['Computers'], dtype=

Here is the screenshot of the stack trace Stack trace