Question

我正在尝试使用flask将Ann churn模型部署到Web应用程序

我从github尝试了这段代码

Service.py

#Import Flask
from flask import Flask, request
import numpy as np
import tensorflow as tf
from keras.models import load_model
from sklearn.externals import joblib

#Initialize the application service
app = Flask(__name__)

#Upload Model
def UploadModel():

    FILENAME_MODEL_TO_LOAD = "ann_model-250-50.h5"
    FILENAME_SCALER_TO_LOAD = "stdScaler.save"
    FILENAME_LABELENCODER_X1_TO_LOAD = "labelEncoder_X_1.save"
    FILENAME_LABELENCODER_X2_TO_LOAD = "labelEncoder_X_2.save"
    MODEL_PATH = "../../../models/classification/churn-clients"

    # Load the RNA from disk
    loaded_model = load_model(MODEL_PATH + "/" + FILENAME_MODEL_TO_LOAD)
    print("Modelo cargado de disco << ", loaded_model)

    # Load the parameters used
    loaded_scaler = joblib.load(MODEL_PATH + "/" + FILENAME_SCALER_TO_LOAD)
    loaded_labelEncoderX1 = joblib.load(MODEL_PATH + "/" + FILENAME_LABELENCODER_X1_TO_LOAD)
    loaded_labelEncoderX2 = joblib.load(MODEL_PATH + "/" + FILENAME_LABELENCODER_X2_TO_LOAD)

    graph = tf.get_default_graph()
    global loaded_model,loaded_scaler,loaded_labelEncoderX1,loaded_labelEncoderX2, graph

#Define a route
@app.route('/abandonment/client/', methods=['GET','POST'])
def default():
    # print (request.data)
    # print (request.args)
    # print (request.form)
    data = None
    if request.method == 'GET':
        print ("GET Method")
        data = request.args

    if request.method == 'POST':
        print ("POST Method")
        if (request.is_json):
            data = request.get_json()

    print("Data received:", data)

    # Getting parameters
    scoreCrediticio = data.get("scoreCrediticio")
    pais = data.get("pais")
    genero = data.get("genero")
    edad = data.get("edad")
    tenencia = data.get("tenencia")
    balance = data.get("balance")
    numDeProductos = data.get("numDeProductos")
    tieneTarjetaCredito = data.get("tieneTarjetaCredito")
    esMiembroActivo = data.get("esMiembroActivo")
    salarioEstimado = data.get("salarioEstimado")

    print ("\nscoreCrediticio: ",scoreCrediticio,
            "\npais: ", pais,
            "\ngenero: ", genero,
            "\nedad: ", edad,
            "\ntenencia: ", tenencia,
            "\nbalance: ", balance,
            "\nnumDeProductos: ", numDeProductos,
            "\ntieneTarjetaCredito: ", tieneTarjetaCredito,
            "\nesMiembroActivo: ", esMiembroActivo,
            "\nsalarioEstimado: ", salarioEstimado)

    # Transformed / Climbing data
    [pais] = loaded_labelEncoderX1.transform([pais])
    [genero] = loaded_labelEncoderX2.transform([genero])

    clients = np.array([scoreCrediticio,pais,genero,edad,tenencia,balance,numDeProductos,tieneTarjetaCredito,esMiembroActivo,salarioEstimado])
    print("clients: ", clients)
    clients = loaded_scaler.transform([clients])
    print("clients Norm: ", clients)

    with graph.as_default():
        result     = ""
        score = loaded_model.predict(clients)
        print("Final score: ", score)
        abandona = (score > 0.5)
        if abandona:
            result += "Abandona"
        else:
            result += "No abandona"
        return result + ', score: ' + str(score[0])

# Run de application
app.run(host='0.0.0.0',port=5000)

Index.html

  <head>
  <meta http-equiv="Content-Type" content="text/html; charset=ISO 8859-1">
  <meta http-equiv="X-UA-Compatible" content="IE=edge">
  <title>Predictor de Abandono de Clientes</title>
  <meta name="description" content="">
  <meta name="viewport" content="width=device-width, initial-scale=1">
  <link rel="stylesheet" href="https://fonts.googleapis.com/css?family=Roboto:300,400,500,700,400italic">
  <link rel="stylesheet" href="https://ajax.googleapis.com/ajax/libs/angular_material/1.1.4/angular-material.min.css">
</head>

<body ng-app="myApp" ng-cloak style="background-color: #efefef" ng-controller="myCtrl" layout="column">
  <md-toolbar>
    <div class="md-toolbar-tools">
      <span>Predictor de abandono de clientes</span>
    </div>
  </md-toolbar>
  <md-content flex style="width:420">
    <md-card>
      <md-card-content>
        <p><em>Ingrese los datos del cliente:</em></p>
       </md-card-content>
    </md-card>
    <md-card>
      <md-card-content>

        <div layout="row" layout-align="space-between center">
          <span>Score crediticio</span>
          <md-slider-container> {: data.scoreCrediticio :}&nbsp;
            <md-slider aria-label="Slider" min="350" max="850" ng-model="data.scoreCrediticio" id="age-slider"></md-slider>
          </md-slider-container>
        </div>

        <div layout="row" layout-align="space-between center">
          <span>País</span>
          <md-select ng-model="data.pais" placeholder="Select">
            <md-option value="France">France</md-option>
            <md-option value="Germany">Germany</md-option>
            <md-option value="Spain">Spain</md-option>
          </md-select>
        </div>

        <div style="margin-bottom:32px" layout="row" layout-align="space-between center">
          <span>Género</span>
          <md-radio-group ng-model="data.genero" layout="row">
            <md-radio-button value="Male">Male</md-radio-button>
            <md-radio-button value="Female">Female</md-radio-button>
          </md-radio-group>
        </div>

        <div layout="row" layout-align="space-between center">
          <span>Edad</span>
          <md-slider-container> {: data.edad :}&nbsp;
            <md-slider aria-label="Slider" min="18" max="92" ng-model="data.edad" id="age-slider"></md-slider>
          </md-slider-container>
        </div>

        <div layout="row" layout-align="space-between center">
          <span>Tenencia</span>
          <md-slider-container> {: data.tenencia :}&nbsp;
            <md-slider aria-label="Slider" min="0" max="10" ng-model="data.tenencia" id="age-slider"></md-slider>
          </md-slider-container>
        </div>

        <div layout="row" layout-align="space-between center">
          <span>Balance</span>
          <input type="number" ng-model="data.balance"></input>
        </div>

        <div style="margin-bottom:32px" layout="row" layout-align="space-between center">
          <span>Número de productos</span>
          <md-select ng-model="data.numDeProductos" placeholder="Select">
            <md-option value="0">0</md-option>
            <md-option value="1">1</md-option>
            <md-option value="2">2</md-option>
            <md-option value="3">3</md-option>
            <md-option value="4">4</md-option>
          </md-select>
        </div>

        <div style="margin-bottom:32px" layout="row" layout-align="space-between center">
          <span>¿Tiene Tarjeta de Crédito?</span>
          <md-radio-group ng-model="data.tieneTarjetaCredito" layout="row">
            <md-radio-button value="1">Sí</md-radio-button>
            <md-radio-button value="0">No</md-radio-button>
          </md-radio-group>
        </div>

        <div style="margin-bottom:32px" layout="row" layout-align="space-between center">
          <span>¿Es miembro activo?</span>
          <md-radio-group ng-model="data.esMiembroActivo" layout="row">
            <md-radio-button value="1">Sí</md-radio-button>
            <md-radio-button value="0">No</md-radio-button>
          </md-radio-group>
        </div>

        <div layout="row" layout-align="space-between center">
          <span>Salario anual</span>
          <input type="number" ng-model="data.salarioEstimado"></input>
        </div>

      </md-card-content>

      <md-card-actions layout="row" layout-align="end center">
        <md-button ng-click="submit()" class="md-accent md-raised">Predecir</md-button>
      </md-card-actions>

    </md-card>
    <md-card>
      <md-card-content>
        <div layout="row" layout-align="space-between center">
          <span style="font-weight:bold">Predicción:</span>
          <span style="color:blue;font-weight:bold">{: result :}</span>
        </div>
      </md-card-content>
    </md-card>
  </md-content>

  <script src="https://ajax.googleapis.com/ajax/libs/angularjs/1.5.3/angular.js"></script>
  <script src="https://ajax.googleapis.com/ajax/libs/angularjs/1.5.3/angular-aria.js"></script>
  <script src="https://ajax.googleapis.com/ajax/libs/angular_material/1.1.4/angular-material.min.js"></script>
  <script src="https://ajax.googleapis.com/ajax/libs/angularjs/1.5.3/angular-animate.js"></script>

  <script>
    var app = angular.module('myApp', ['ngMaterial']);
    app.config(['$interpolateProvider', function($interpolateProvider) {
      $interpolateProvider.startSymbol('{:');
      $interpolateProvider.endSymbol(':}');
    }]);

    app.config(function($mdThemingProvider) {
      $mdThemingProvider.theme('default')
        .primaryPalette('blue-grey')
        .accentPalette('orange');
    });

    app.controller('myCtrl', ['$scope', '$http',
      function($scope, $http) {
        $scope.data = {};

        $scope.submit = function() {
          console.log($scope.data)
          $scope.result = "Wait...";

          $http({
            method : 'POST',
            url : 'http://localhost:5000/abandono/cliente/',
            headers: {
                'Content-Type': 'application/json'
            },
            data: $scope.data
          }).success(function(data, status, headers, config) {
            $scope.result = data;

          }).error(function(data, status, headers, config) {
            $scope.result = 'Error';

          });

        }
      }
    ]);
  </script>
</body>
</html>

，当我运行它并在浏览器中打开http://localhost:5000/abandonment/client/时，找不到服务器上找不到请求的URL。如果您手动输入网址，请检查拼写，然后重试。

如何使用flask部署Ann churn模型？

0 个答案: