Skip to content

KaJunho/bedrock-data-analysis-agent

BranchesTags

Repository files navigation

Bedrock Data Analysis Agent

A data analysis assistant for video game sales and reviews data analysis based on Bedrock Agent, developed upon video_games_sales_assistant_with_amazon_bedrock_agents. For video demo, check out demo.

Architecture

The architecture of this project is shown below. new-agent-arch

Backend Deployment

Prerequisites

Before you begin, ensure you have:

aws iam create-service-linked-role --aws-service-name rds.amazonaws.com

Deploy the Back-End Services with AWS SAM

Navigate to the SAM project folder (sam-bedrock-video-games-sales-assistant/) and execute::

sam build

Caution

If you encounter a Build Failed error, you might need to change the Python version in the template.yaml file. By default, the Lambda function uses Python 3.13. You can modify this setting on line 121 of the template.yaml file to use a Python version lower than 3.13 that you have installed.

Now deploy the SAM application under the SAM project foler (sam-bedrock-video-games-sales-assistant/):

sam deploy --guided --capabilities CAPABILITY_NAMED_IAM

If not the first time deployment and do not modify config file, use this:

sam deploy --capabilities CAPABILITY_NAMED_IAM

Use the following value arguments for the deployment configuration:

  • Stack Name : sam-bedrock-video-games-sales-assistant
  • AWS Region : us-east-1
  • Parameter PostgreSQLDatabaseName : video_games_sales
  • Parameter AuroraMaxCapacity : 2
  • Parameter AuroraMinCapacity : 1
  • Parameter S3BucketName : game-reviews-0630
  • Parameter KnowledgeBaseName : game-kb-0630
  • Parameter KnowledgeBaseDescription : Game reviews knowledge base
  • Parameter DataSourceName : game_reviews-0630
  • Parameter DataSourceDescription : Data Source for Amazon DA Assistant Knowledge Base
  • Confirm changes before deploy : Y
  • Allow SAM CLI IAM role creation : Y
  • Disable rollback : N
  • Save arguments to configuration file : Y
  • SAM configuration file : samconfig.toml
  • SAM configuration environment : default

After the SAM project preparation and changeset created, confirm the following to start the deployment:

  • Deploy this changeset? [y/N]: Y

After deployment completes, the following services will be created:

Note

To learn about agent creation configuration, please refer to this tutorial, which provides step-by-step guidance for setting up an Amazon Bedrock Agent in the AWS Console.

Load Sample Data into Aurora PostgreSQL Database

Set up the required environment variables:

# Set the stack name environment variable
export STACK_NAME=sam-bedrock-video-games-sales-assistant

# Retrieve the output values and store them in environment variables
export SECRET_ARN=$(aws cloudformation describe-stacks --stack-name "$STACK_NAME" --query "Stacks[0].Outputs[?OutputKey=='SecretARN'].OutputValue" --output text)
export DATA_SOURCE_BUCKET_NAME=$(aws cloudformation describe-stacks --stack-name "$STACK_NAME" --query "Stacks[0].Outputs[?OutputKey=='DataSourceBucketName'].OutputValue" --output text)
export AURORA_SERVERLESS_DB_CLUSTER_ARN=$(aws cloudformation describe-stacks --stack-name "$STACK_NAME" --query "Stacks[0].Outputs[?OutputKey=='AuroraServerlessDBClusterArn'].OutputValue" --output text)
cat << EOF
STACK_NAME: ${STACK_NAME}
SECRET_ARN: ${SECRET_ARN}
DATA_SOURCE_BUCKET_NAME: ${DATA_SOURCE_BUCKET_NAME}
AURORA_SERVERLESS_DB_CLUSTER_ARN: ${AURORA_SERVERLESS_DB_CLUSTER_ARN}
EOF

Execute the following command to create the database and load the sample data, under the SAM project foler (sam-bedrock-video-games-sales-assistant/).

python3 resources/create-sales-database.py

The script uses the [video_games_sales_random_avg_sales.csv] as the data source.

Sync Knowledge Base Data Source

Navigate to your Amazon Bedrock Knowledge Bases control panel:

  • Click game-kb-0630
  • In Data source section, check the data source game_reviews-0630
  • Click Sync button on the top right

Test the Agent in AWS Console

Navigate to your Amazon Bedrock Agent named video-game-sales-data-analyst:

  • Click Edit Agent Builder
  • In the Agent builder section click Save
  • Click Prepare
  • Click Test

Try these sample questions in Chinese:

  • 你好
  • 请找出总销量最佳的三款游戏。
  • 2015年至2019年发布的游戏，在各地区的总销售额是多少？请以百分比形式给我数据。
  • 请简单介绍一下Moon Storm这款游戏

Create Agent Alias for Front-End Application

To use the agent in your front-end application:

  • Go to your Agent Overview
  • Click Create Alias

You can now proceed to the next section. The tutorial will ask you for your Agent Alias along with the other services that you have created so far.

Frontend Deployment

Prerequisites

Before you begin, ensure you have:

  • An Alias created for your Amazon Bedrock Agent from the Generative AI Application - Data Source and Amazon Bedrock Agent Deployment tutorial
  • Node.js version 18+
  • React Scripts installed:
npm install react-scripts

Set Up the Front-End Application

Navigate to the React application folder (amplify-video-games-sales-assistant-sample/) and install the Reac application dependencies:

npm install

Configure IAM User Access for Front-End Permissions

Update the values with your <agent_arn>, <agent_id>, <account_id> and <question_answers_table_arn> that you can find in the outputs from the SAM tutorial.

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Sid": "InvokeBedrockAgent",
            "Effect": "Allow",
            "Action": [
                "bedrock:InvokeAgent"
            ],
            "Resource": [
                "<agent_arn>",
                "arn:aws:bedrock:*:<account_id>:agent-alias/<agent_id>/*"
            ]
        },
        {
            "Sid": "InvokeBedrockModel",
            "Effect": "Allow",
            "Action": [
                "bedrock:InvokeModel"
            ],
            "Resource": [
                "arn:aws:bedrock:*:<account_id>:inference-profile/us.anthropic.claude-3-5-sonnet-20241022-v2:0",
                "arn:aws:bedrock:us-east-2::foundation-model/anthropic.claude-3-5-sonnet-20241022-v2:0",
                "arn:aws:bedrock:us-east-1::foundation-model/anthropic.claude-3-5-sonnet-20241022-v2:0",
                "arn:aws:bedrock:us-west-2::foundation-model/anthropic.claude-3-5-sonnet-20241022-v2:0"
            ]
        },
        {
            "Sid": "DynamoDB",
            "Effect": "Allow",
            "Action": [
                "dynamodb:Query"
            ],
            "Resource": "<question_answers_table_arn>"
        }
    ]
}

Configure Environment Variables

  • Rename the file src/sample.env.js to src/env.js and update the following environment variables:

    • AWS Credentials and Region:

      • ACCESS_KEY_ID
      • SECRET_ACCESS_KEY
      • AWS_REGION

    • Agent and table information that you can find in the CloudFormation Outputs from the SAM project:

      • AGENT_ID
      • AGENT_ALIAS_ID
      • QUESTION_ANSWERS_TABLE_NAME

    • Also, you can update the general application description:

      • APP_NAME
      • APP_SUBJECT
      • WELCOME_MESSAGE

Test Your Data Analyst Assistant

Start the application locally:

npm start

Test questions

Here are some test questions in Chinese.

你好！
你能如何帮助我？
2015-2025年哪一年发布的游戏数量最多？
请找出总销量最佳的三款游戏。
2015年至2019年发布的游戏，在各地区的总销售额是多少？请以百分比形式给我数据。
请简单介绍一下Moon Storm这款游戏
请列出玩家们对Moon Storm这款游戏的正面评价（包含玩家姓名ID及邮箱）
请总结玩家们对Moon Storm这款游戏的负面评价（包含玩家姓名ID及邮箱）
请找出2015年至2025年期间全球最畅销的游戏产品，以饼状图的形式展示该最畅销游戏产品在不同地区的销售数据分布情况。然后，列举玩家们（包含玩家的个人资料信息）对这款游戏的评价。最后基于这些评价简要解释这款游戏能够在全球范围内最畅销的原因。

Thank You

License

This project is licensed under the Apache-2.0 License.

About

A data analysis assistant for video game sales&reviews data analysis based on Bedrock Agent.

Resources

Readme
Activity

Stars

2 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

 
 
 

Contributors

Languages