Income Prediction

Description of the problem

The goal of this project is a simple one: when provided with data such as occupation, country, and age, can we predict income? Specifically, can we predict if a person's income will be over $50k or not?

The uses for this kind of data are endless, but here are a few:

• Marketing - target the higher or lower income group, whatever matches your brand needs
• Finance - use as a factor to determine lending terms or detect fraud
• Politics - use to help decide where you should be focusing campaign efforts
• Outreach - determine underserved areas and provide assistance
• Housing - locate areas where affordable housing should be built
  
  

Instructions on how to run the project

Get the code

In your terminal:

1. Download repo: git clone https://github.com/ANKershaw/Income_Prediction.git
2. Navigate to project folder:cd Income_Prediction. All future commands assume you're in this folder.
   
   

Dependency Management

1. Install uv, following the instructions here
2. Install packages: uv sync --locked
   
   

If you want to test the deployed service

I've provided a script to do this. You can run it via:
uv run income_prediction.py .

The service is currently deployed here: https://damp-field-1507.fly.dev/predict
The swagger page is here: https://damp-field-1507.fly.dev/docs
The service will stay running until January 1, 2026, unless fly.io becomes prohibitively expensive.

If you want to deploy the docker container locally

1. Build the container docker build -t income-prediction
2. Run the service docker run -it --rm -p 9696:9696 income-prediction
   
   

How to test a local deploy

Browser

In your browser, navigate to http://0.0.0.0:9696/docs and use the swagger interface with this data:

{
    "age": 23,
    "workclass": "Private",
    "education": "Some-college",
    "marital-status": "Never-married",
    "occupation": "Handlers-cleaners",
    "race": "White",
    "sex": "Female",
    "capital-gain": 0,
    "capital-loss": 0,
    "hours-per-week": 18,
    "native-country": "United-States"
}

via Curl

here is the curl command:

curl -X 'POST' \
  'http://0.0.0.0:9696/predict' \
  -H 'accept: application/json' \
  -H 'Content-Type: application/json' \
  -d '{
  "age": 17,
  "workclass": "State-gov",
  "education": "Bachelors",
  "marital-status": "Never-married",
  "occupation": "Adm-clerical",
  "race": "White",
  "sex": "Male",
  "capital-gain": 99999,
  "capital-loss": 4356,
  "hours-per-week": 1,
  "native-country": "United-States"
}'

Data Source

The dataset is from the UC Irvine Machine Learning Repository. The link is here: https://archive.ics.uci.edu/dataset/2/adult. The data is pulled from the 1994 Census database. With inflation, the $50,000 benchmark is around $111,000 today.

Context for this Project

This repo is the midterm project for DataTalks Club's Machine Learning Zoomcamp. This repo is based around the specifications and requirements listed here: https://github.com/DataTalksClub/machine-learning-zoomcamp/tree/master/projects.