|
| 1 | +{ |
| 2 | + "cells": [ |
| 3 | + { |
| 4 | + "cell_type": "markdown", |
| 5 | + "metadata": {}, |
| 6 | + "source": [ |
| 7 | + "# Putting everything together" |
| 8 | + ] |
| 9 | + }, |
| 10 | + { |
| 11 | + "cell_type": "markdown", |
| 12 | + "metadata": {}, |
| 13 | + "source": [ |
| 14 | + "# Build our deployment package\n", |
| 15 | + "\n", |
| 16 | + "* We need to start a EC2 instance with AMI (at least 2GB of memory)" |
| 17 | + ] |
| 18 | + }, |
| 19 | + { |
| 20 | + "cell_type": "raw", |
| 21 | + "metadata": { |
| 22 | + "collapsed": false |
| 23 | + }, |
| 24 | + "source": [ |
| 25 | + "# Connect to your instance\n", |
| 26 | + "\n", |
| 27 | + "sudo yum -y update\n", |
| 28 | + "sudo yum -y install gcc-c++ python27-devel atlas-sse3-devel lapack-devel\n", |
| 29 | + "\n", |
| 30 | + "sudo /usr/local/bin/pip install -U pip\n", |
| 31 | + "sudo /usr/local/bin/pip install tweepy numpy scipy scikit-learn\n", |
| 32 | + "\n", |
| 33 | + "cd /usr/local/lib/python2.7/site-packages/\n", |
| 34 | + "zip -r9 ~/MyLambdaPackage.zip *\n", |
| 35 | + "cd /usr/local/lib64/python2.7/site-packages/\n", |
| 36 | + "zip -r9 ~/MyLambdaPackage.zip *\n", |
| 37 | + "\n", |
| 38 | + "# You can download your package now an put in on S3\n", |
| 39 | + "# And then strip everything....." |
| 40 | + ] |
| 41 | + }, |
| 42 | + { |
| 43 | + "cell_type": "raw", |
| 44 | + "metadata": { |
| 45 | + "collapsed": false |
| 46 | + }, |
| 47 | + "source": [ |
| 48 | + "https://github.com/ryansb/sklearn-build-lambda" |
| 49 | + ] |
| 50 | + }, |
| 51 | + { |
| 52 | + "cell_type": "markdown", |
| 53 | + "metadata": {}, |
| 54 | + "source": [ |
| 55 | + "# 1) Create the predict.py \n", |
| 56 | + "\n", |
| 57 | + "This function will receive a stream from S3, it will load the data and predict the sentiment and then save the json file in s3." |
| 58 | + ] |
| 59 | + }, |
| 60 | + { |
| 61 | + "cell_type": "markdown", |
| 62 | + "metadata": {}, |
| 63 | + "source": [ |
| 64 | + "# 2) Create the reader.py\n", |
| 65 | + "\n", |
| 66 | + "This function will be called every 1 min from AWS CloudWatch Events and it will save a list of 50 tweets in S3" |
| 67 | + ] |
| 68 | + }, |
| 69 | + { |
| 70 | + "cell_type": "markdown", |
| 71 | + "metadata": {}, |
| 72 | + "source": [ |
| 73 | + "# 3) Create the buckets\n", |
| 74 | + "You will need 2 buckets:\n", |
| 75 | + " * twitter.reader\n", |
| 76 | + " * twitter.sentiment\n", |
| 77 | + " * twitter.deploypack" |
| 78 | + ] |
| 79 | + }, |
| 80 | + { |
| 81 | + "cell_type": "markdown", |
| 82 | + "metadata": {}, |
| 83 | + "source": [ |
| 84 | + "# 4) Create the deployment package\n", |
| 85 | + "\n", |
| 86 | + "* cp deployment_pack.zip deployment_pack_with_code.zip\n", |
| 87 | + "\n", |
| 88 | + "* zip -9r deployment_pack_with_code.zip reader.py predict.py models/clf.pkl models/vocabulary.pkl" |
| 89 | + ] |
| 90 | + }, |
| 91 | + { |
| 92 | + "cell_type": "markdown", |
| 93 | + "metadata": {}, |
| 94 | + "source": [ |
| 95 | + "# 5) Upload everything\n", |
| 96 | + "* load the deploypack in twitter.deploypack\n", |
| 97 | + "* create a lambda function (call it \"predict\") with S3 as a trigger, put twitter.reader as a bucket to control\n", |
| 98 | + "* create a lambda function (call it \"reader\") with CloudWatch\n", |
| 99 | + "* for both functions set 256MB as memory and 60 secs as timeout\n", |
| 100 | + "* Watch your twitter.sentiment !" |
| 101 | + ] |
| 102 | + }, |
| 103 | + { |
| 104 | + "cell_type": "markdown", |
| 105 | + "metadata": {}, |
| 106 | + "source": [ |
| 107 | + "# Error using pickle with the vectorizer" |
| 108 | + ] |
| 109 | + }, |
| 110 | + { |
| 111 | + "cell_type": "markdown", |
| 112 | + "metadata": {}, |
| 113 | + "source": [ |
| 114 | + "Probably related to Pickle and sparse matrix" |
| 115 | + ] |
| 116 | + }, |
| 117 | + { |
| 118 | + "cell_type": "markdown", |
| 119 | + "metadata": {}, |
| 120 | + "source": [ |
| 121 | + "format not found: AttributeError\n", |
| 122 | + "Traceback (most recent call last):\n", |
| 123 | + "File \"/var/task/predict.py\", line 32, in handler\n", |
| 124 | + "tweets_vectors = vectorizer.transform([\"Hillary doesn't got what it takes...Trump neither but Hillary my God! https://t.co/7OOeEO4Epi\"])\n", |
| 125 | + "File \"/var/task/sklearn/feature_extraction/text.py\", line 1334, in transform\n", |
| 126 | + "return self._tfidf.transform(X, copy=False)\n", |
| 127 | + "File \"/var/task/sklearn/feature_extraction/text.py\", line 1037, in transform\n", |
| 128 | + "X = X * self._idf_diag\n", |
| 129 | + "File \"/var/task/scipy/sparse/base.py\", line 319, in __mul__\n", |
| 130 | + "return self._mul_sparse_matrix(other)\n", |
| 131 | + "File \"/var/task/scipy/sparse/compressed.py\", line 478, in _mul_sparse_matrix\n", |
| 132 | + "other = self.__class__(other) # convert to this format\n", |
| 133 | + "File \"/var/task/scipy/sparse/compressed.py\", line 28, in __init__\n", |
| 134 | + "if arg1.format == self.format and copy:\n", |
| 135 | + "File \"/var/task/scipy/sparse/base.py\", line 525, in __getattr__\n", |
| 136 | + "raise AttributeError(attr + \" not found\")\n", |
| 137 | + "AttributeError: format not found" |
| 138 | + ] |
| 139 | + } |
| 140 | + ], |
| 141 | + "metadata": { |
| 142 | + "kernelspec": { |
| 143 | + "display_name": "Python 2", |
| 144 | + "language": "python", |
| 145 | + "name": "python2" |
| 146 | + }, |
| 147 | + "language_info": { |
| 148 | + "codemirror_mode": { |
| 149 | + "name": "ipython", |
| 150 | + "version": 2 |
| 151 | + }, |
| 152 | + "file_extension": ".py", |
| 153 | + "mimetype": "text/x-python", |
| 154 | + "name": "python", |
| 155 | + "nbconvert_exporter": "python", |
| 156 | + "pygments_lexer": "ipython2", |
| 157 | + "version": "2.7.12" |
| 158 | + } |
| 159 | + }, |
| 160 | + "nbformat": 4, |
| 161 | + "nbformat_minor": 1 |
| 162 | +} |
0 commit comments