RoaldSchuring

    wine_name_lookup = client.get_object(Bucket='data-science-wine-reviews',
                                         Key='nearest_neighbors/data/wine_reviews_select_cols.csv')
    wine_name_lookup = pd.read_csv(wine_name_lookup['Body'])

    recommendation_indices = list(result[1])
    recommendation_indices = [int(n) for n in recommendation_indices]

    recommendations = []
    for i in recommendation_indices:
        suggested_wine = wine_name_lookup.at[i, 'Name']

RoaldSchuring

response = runtime.invoke_endpoint(EndpointName='sagemaker-scikit-learn-2019-07-04-13-00-07-919',
                                   ContentType='application/json',
                                   Body=wine_vector_output)

def decode(s, encoding="ascii", errors="ignore"):
    return s.decode(encoding=encoding, errors=errors)

result = json.loads(decode(response['Body'].read()))

RoaldSchuring

wine_vector = sum(word_vectors) / len(word_vectors)
wine_vector_output = json.dumps(wine_vector.tolist())

RoaldSchuring

    obj = client.get_object(Bucket='data-science-wine-reviews', Key='word_vectors_idf.csv')
    wine_df = pd.read_csv(obj['Body'])
    wine_df.set_index(['word'], inplace=True)

    word_vectors = []
    for p in payload:
        word_vector_string = wine_df.at[p, 'word_vec_idf']
        word_vector_string = word_vector_string.replace('[', '').replace(r'\n', '').replace(']', '')
        word_vector = np.fromstring(word_vector_string, dtype=float, sep='  ')
        word_vectors.append(word_vector)

RoaldSchuring

import json
import boto3
import pandas as pd
import numpy as np
from six import BytesIO

def lambda_handler(event, context):
    client = boto3.client('s3')
    runtime = boto3.client('runtime.sagemaker')

RoaldSchuring

from sklearn.preprocessing import normalize

# open the vectors.txt file containing all the trained word embeddings, extracting the descriptors & embeddings
num_points = len(open('vectors.txt','r').read().split('\n'))

first_line = True
index_to_word = []
with open("vectors.txt","r") as f:
    for line_num, line in enumerate(f):
        if first_line:

RoaldSchuring

s3 = boto3.resource('s3')
key = bt_model.model_data[bt_model.model_data.find("/", 5)+1:]
s3.Bucket(bucket).download_file(key, 'model.tar.gz')

!tar -xvzf model.tar.gz

RoaldSchuring

train_data = sagemaker.session.s3_input(train_data, distribution='FullyReplicated', 
                        content_type='text/plain', s3_data_type='S3Prefix')
data_channels = {'train': train_data}

bt_model.fit(inputs=data_channels, logs=True)

RoaldSchuring

sess = sagemaker.Session()

# define the specifications of the sagemaker training instance
bt_model = sagemaker.estimator.Estimator(container,
                                         role, 
                                         train_instance_count=2, 
                                         train_instance_type='ml.c4.2xlarge',
                                         train_volume_size = 5,
                                         train_max_run = 360000,
                                         input_mode= 'File',

RoaldSchuring

train_data = 's3://{}/wine-corpus.txt'.format(bucket)
s3_output_location = 's3://{}/output'.format(bucket)

region_name = boto3.Session().region_name
container = sagemaker.amazon.amazon_estimator.get_image_uri(region_name, "blazingtext", "latest")
print('Using SageMaker BlazingText container: {} ({})'.format(container, region_name))