SlapsRoofOfCar.bot

"Car Salesman: slaps roof of car this baby here can fit so many predictive models OwO"

The objective of this bot is to predict the value/worth-it factor of buying a used car given the parameters: year, price, and mileage (of the used car). By using neural networks with a dataset of roughly 200 slices, the bot will be able to tell you if the car is worth it based on current market trends for that specific car.

TLDR; Found the perfect used car? Use this bot to check if it's actually a good deal. Goodluck!

*Note: location and prices only in the US, limitation of the dataset.

Some examples of the model training & graphs of the results

Model training:

Model Loss & Accuracy graphs:

Getting Started

These instructions will get you a copy of the project up and running on your local machine for development and testing purposes. See deployment for notes on how to deploy the project on a live system.

Prerequisites

1. Pulling datasets from http://myslu.stlawu.edu/~clee/dataset/autotrader/
2. Preprocessing this CSV dataset into (X, Y) (210, 2) (45, 2) (45, 2) (210, 1) (45, 1) (45, 1)

• Training set
• Validation set
• Test set

Installing

1. Requires the following packages to be installed onto your system

• sklearn
• pandas
• numpy
• keras
• matplotlib

Running the tests

1. (Optional)

• Use venv: source bin/activate

2. To pull the dataset & preprocess the data

• python3 main.py with the flags --radius, --search_results, --without_csv, --dry_run
• Use --help if you don't know what the flags represent!

3. The script will prompt you to enter in the details of the car you're currently interested in

• Ex. kia, forte, 2017, 10.5 (thousand), 18.5 (thousand), 32703 (zipcode in Florida)

4. Give the model some time to evaluate. The result should be printed after a minute or so: Under evaluation or Over evaluation

Agenda

1. Create a web interface to run the simulation
2. Reduce excessive deprecated logging from tensorflow

Acknowledgments

• https://www.freecodecamp.org/news/how-to-build-your-first-neural-network-to-predict-house-prices-with-keras-f8db83049159/