samtowne/basketballshootinginsights

Measure basketball shooting drill results over time.

Basketball Shooting Drill Data Processing and Visualization 🏀

Provides insight into basketball shooting drill results over time.

App Strategy 🏗️

• Event-driven and decoupled
• Serverless data processing
• Low cost per handled event
• Data Visualization with pretty colors

Tech Toolbox 🧰

• Terraform
• API Gateway, Lambda, DyanmoDB, S3, Athena, Glue, Simple Email Service, IAM, Quicksight
• Hashicorp Configuration Language, Python, Node.js
• Google Forms

Data Visualization

Data visualization is performed using Amazon Quicksight. The processed data is read using Quicksight's SPICE engine to provide visualizations. SPICE is able to determine the schema of JSON files stored in an S3 bucket and this made for a quick visualization of the collected data. A disadvantage of the SPICE engine is that it performs an entire read of the dataset each time a refresh is required.

Application Flow

Submit ➡️ Collect ➡️ Pre-Process ➡️ Process ➡️ Respond ➡️ Clean

Submit

• Google Form submission triggers an HTTP post to an AWS API Gateway.
• The API Gateway invokes Collect Lambda.

Collect

• Collect stores the data to an S3 bucket, creates a temp file for this event, and invokes Pre-Process Lambda.

Pre-Process

• Pre-Process executes an Athena Query to create a database and glue table for this event.
• When the query completes, a bucket notification invokes Process Lambda.

Process

• Process executes another Athena Query. This is to obtain the data needed to perform average shooting percentage calculations against all previous submissions of this drill type.
• When the query completes, a bucket notification invokes Respond Lambda.

Respond

• Respond sends an email with the results and invokes Clean.

Clean

• Clean runs to remove temp files.

The response email provides drill stats for immediate feedback of the processed drill, and compares against historical submissions.

Presentation

• The data is ingested to Amazon Quicksight to provide dashboard summary views.

Terraform Modules

• Bootstrap creates the Terraform state bucket and dynamodb state-locking table.
• Collection feeds data into the application.
• Processing draws insight from the data. The collection step results in raw json objects stored in s3. Each POST to the api gateway results in an individual json object being stored. Processing runs queries against the recent data as well as the historicall (previously submitted) json files.
• Response emails the results to the user.
• Cleanup deletes temporary files created during runtime.
• Presentation configuration for Amazon Quicksight dashboards.

Shooting Drills

Each drill involves attempting 4 shots from each of 11 locations. The data is segregated by API route such as https://AWS-Generated-ID.execute-api.AWS-Region.amazonaws.com/threepoint or https://AWS-Generated-ID.execute-api.AWS-Region.amazonaws.com/midrange.

Three Point

Mid Range