Welcome to the portfolio
Back to projects

Nexo · Applied machine learning

Credit assessment system with Machine Learning

Nexo demonstration.
Context
Personal project
Contribution
Full stack development

Problem

Make the assessment of financial profiles understandable and explore how changing a variable affects an estimate.

Solution

I built individual assessments and batches of up to ten applications, scenario comparison, Excel exports and a dashboard updated with the current session’s queries.

Architecture

Django MVT monolith. A Random Forest model and data processing with Pandas, NumPy and scikit-learn. Deployed with Gunicorn, WhiteNoise and Docker.

Step 01 / 04 · Browser

The form submits the fictional “Caso Lima” financial profile.

Read the walkthrough
  1. The form submits the fictional “Caso Lima” financial profile.
  2. Django prepares the application variables for the model.
  3. Random Forest estimates an 83.82% approval probability for this profile.
  4. The dashboard adds the assessment to the results for this session.
DeploymentRenderDockerGunicornWhiteNoise

Technologies

Application

  • Python
  • Django
  • Django Templates
  • HTML
  • CSS
  • JavaScript

Model

  • scikit-learn
  • Random Forest
  • StandardScaler
  • Pandas
  • NumPy

Delivery

  • OpenPyXL
  • Gunicorn
  • WhiteNoise
  • Docker
  • Render