Skip to content

full-stack web app that scores cars using neural network trained on UCI car evaluation dataset

License

Notifications You must be signed in to change notification settings

FrederickRoman/Carscore

Repository files navigation

Carscore

Carscore logo

ML-drive car resell scoring

This is a scoring system that uses a neural network train on UCI car evaluation dataset. This neural net runs on the back-end and send its response through a REST API.

Carscore home page mobile mockup Carscore home page desktop mockup

Live website

See Carscore website.

Tech stack used in this project (all is in this repo)

  • Client-side:
    • Vue.js (2)
    • Vuetify
  • Server-side:
    • Node
    • Express
  • Machine Learning (NN):
    • Tensorflow.js (Node)

Project setup

npm install

Compiles and hot-reloads for client development

npm run client_dev

Compiles and hot-reloads for server development

npm run server_dev

Compiles and minifies for production

npm run postinstall

Lints and fixes client-side files

npm run client_lint

Retrains neural network (this is will overwrite the current one). The best one found so far is on the top level as /neural_network.

npm run server_train_nn

Citation (Data)

(a) Creator: Marko Bohanec (b) Donors: Marko Bohanec ([email protected]) Blaz Zupan ([email protected]) (c) Date: June, 1997