Table of Contents

• About the project
• Clustering visualization example
• Build and run guidance
• Generate data
• Clusterize data
• Azure deployment

About the project

Service, that implements clustering technique using K-means algorithm.

Clustering visualization example

The following screenshots show the result of algorithm's execution over the set of 2D points that could be separated into 4 clusters.

Build and run guidance

1. Clone project: git clone https://github.com/SergiySobolev/kmeans-clustering-service.git
2. Go to the root of the project directory: cd kmeans-clustering-service
3. Build backend: ./gradlew build
4. Run backend: java -jar build/libs/kmeans-clustering-service.jar . By default, backend service will start on port 11111.

Generate data

Backend provides endpoint to generate synthetic data that could be divided into clusters:

URL /generatedata

METHOD POST

HEADERS Content-Type: application/json

DATA PARAMS

clusterNum: int,

bounds: 2d array of ints

RESPONSE

data: 2d array of ints

EXAMPLE

Generate data that could be divided into 3 clusters

Request:

POST http://{host}:11111/generatedata

Content-Type: application/json

{"clusterNum": 3, "bounds": [[100, 200],[100, 200],[350, 550],[350, 550],[2000, 2500],[2000, 2500]]}

Response visualization:

Clusterize data

Backend provides endpoint to separate points into clusters:

URL /clusterdata

METHOD POST

HEADERS Content-Type: application/json

DATA PARAMS

type: Algorithm type. Only possible value for the moment is "KMEANS". Another algorithms will be added furter

clusterNum: Number of clusters the data must be divided to

data: Array of 2d point to clusterize

RESPONSE

data: array of clusterized points, where clusterized points is [x,y,cluster_index]

EXAMPLE

Divide data into 3 clusters

Request:

POST http://{host}:11111/clusterdata

Content-Type: application/json

{'type':'KMEANS', 'clusterNum': 3, 'data':generated_data} where generated_data is the result of Generate data request

Response visualization:

Azure deployment

The entire solution can be deployed on Azure Cloud using Container Instances

Prerequisites:

1. Get Microsoft Azure Subscription
2. Install Terraform

Steps:

1. git clone https://github.com/SergiySobolev/kmeans-clustering-service.git
2. cd kmeans-clustering-service/azureiac
3. terraform init
4. terraform plan -target=module.backend_container
5. terraform apply -target=module.backend_container
6. terraform plan -target=module.frontend_container
7. terraform apply -target=module.frontend_container
8. Go to Jupyter Notebook