nycity_taxi_ad_api

Taxi demo with Ad API

Introduction

This document contains the steps required to configure, build and deploy coupon service (Ad Service) into GAE

Prerequisite

1. Make sure Java 8 is installed and JAVA_HOME is set in environment

2. Make sure maven is installed and added in PATH

3. Make sure gcloud is installed and project is set to **"**billion-taxi-rides"

Coupon Service Setup

1. Open terminal or command prompt and execute "**git clone ****https://github.com/springml/nycity_taxi_ad_ap**i"

2. cd nycity_taxi_ad_api/nycity_taxi_ad_api_java

3. To build coupon service execute "mvn clean package"

4. To run coupon service locally execute "java -jar target/nycity-taxi-ad-api-1.0.0.jar"

5. To deploy coupon service into GAE execute "mvn appengine:deploy". Rest service can be accessed using https://coupon-service-dot-billion-taxi-rides.appspot.com/getCoupon

Configuration

1. By default service is set to **coupon-service. **You can change it in /nycity_taxi_ad_api/nycity_taxi_ad_api_java/src/main/appengine/app.yaml

2. Additional configurations related to GAE environment has to be done in /nycity_taxi_ad_api/nycity_taxi_ad_api_java/src/main/appengine/app.yaml

3. By default coupon-service uses CloudML from **billion-taxi-rides. **This can be changed by providing the CloudML REST url in **cloudML.predict.rest.url **property in /nycity_taxi_ad_api/nycity_taxi_ad_api_java/src/main/resources/application.properties

4. For any changes in coupon discount mapping, edit the file present in /nycity_taxi_ad_api/nycity_taxi_ad_api_java/src/main/resources/coupon-discount.properties

**Note:- **Any changes in configuration will be reflected only when the app is rebuilt and redeployed in to GAE.

Invoke Coupon Service

1. Install and open any REST client like Postman or Advanced REST Client…

2. Use the POST method

3. Use the URL https://coupon-service-dot-billion-taxi-rides.appspot.com/getCoupon. Please make sure to change the service is same as you’ve provided in app.yaml

4. Add the "Content-Type" header with “application/json” as value

5. Use the below json content as body

{

• "passenger_count":2,*

• "tpep_pickup_datetime":"2015-01-04 20:06:55",*

• "pickup_latitude":"40.751056671142578",*

• "pickup_longitude":"-73.994209289550781",*

• "dropoff_latitude":"40.749996185302734",*

• "dropoff_longitude":"-73.97900390625"*

}

6. On executing the above request you will get a response like below

{

• "couponId": 9,*

• "discount": "12%"*

}

Run Simulator

Injector

1. Login into Google Cloud Console and select Google Compute Engine from the menu

2. Click on the "SSH" dropdown adjacent to “coupon-service-injector” VM and click “Open in browser window”

3. In terminal switch to root user using sudo su -

4. cd nycity_taxi_ad_api/injector/

5. Edit config_without_docker.sh if you need to increase the speed or change any other configuration

6. Execute source config_without_docker.sh

7. Execute *go run .go

Dataflow

1. Login into Google Cloud Console and select Google Compute Engine from the menu

2. Click on the "SSH" dropdown adjacent to “ coupon-service-simulator” VM and click “Open in browser window”

3. In terminal switch to root user using sudo su -

4. cd dataflow/nycity_taxi_ad_api/coupon_service_simulator/

5. Execute mvn clean compile exec:java -Dexec.mainClass=com.springml.coupon.service.simulator.CouponServiceGDF -e -Dexec.args="--project=billion-taxi-rides --stagingLocation=gs://sml/staging --runner=DataflowPipelineRunner --streaming=true --numWorkers=20 --zone=us-west1-b --couponServiceUrl=https://demo5-test.apigee.net/v1/coupon --apiKey=API_KEY"

Scheduler

1. Login into Google Cloud Console and select Google Compute Engine from the menu

2. Click on the "SSH" dropdown adjacent to “ coupon-service-simulator” VM and click “Open in browser window”

3. In terminal switch to root user using sudo su -

4. cd scheduler-demo/nycity_taxi_ad_api/nycity_taxi_ad_scheduler/

5. Execute java -jar target/nycity-taxi-ad-scheduler-1.0-SNAPSHOT.jar

Run Taxi Dataflow which will publish data into BigQuery

We have to start injector and dataflow.

Dataflow

To start dataflow follow below steps

1. Start sml-tax-dataflow compute-engine and ssh into it

2. sudo su -

3. cd work/dataflow_nyc_taxis_next17_visupipe/dataflow/

4. mvn clean compile exec:java -Dexec.mainClass=com.springml.TaxiGDF -e -Dexec.args="--project=billion-taxi-rides --sinkProject=billion-taxi-rides --stagingLocation=gs://billion-taxi-rides/df-staging/ --runner=DataflowPipelineRunner --streaming=true --numWorkers=1 --zone=us-west1-b"

5. Check whether the stop successfully started by navigating to https://console.cloud.google.com/dataflow?project=billion-taxi-rides

Injector

To start injector follow below steps

1. Start next17keynote-injector compute-engine and ssh into it

2. sudo su -

3. cd nycity_taxi_ad_api/injector/

4. source config_without_docker.bash

5. go run *.go

Contact

SpringML contact: mayank.sharma@springml.com

Google Contact : alankho@google.com

link

Link to Presentation : https://docs.google.com/presentation/d/1Yc7eZ-CkHmsyxWAOHHGFt5q_llA5nyYv1vWlvbjeA_g/edit#slide=id.g2072fb2c8a_0_0

Link to Demo Script : https://docs.google.com/document/d/1xvFpnJTPJKtdI2JSr6RHzQDviPnfEcA-oNa5Tr_QaS8/edit#