/GEPrediction-OSRS

Prediction of Grand Exchange prices with Recurrent Neural Networks

Primary LanguageJupyter Notebook

OSRS Grand Exchange Price Prediction

Start-to-end project where we attempt to harness the power of machine learning to predict Old-school Runescape Grand Exchange prices.

Getting started

The journey so far has been documented in a series of Youtube videos found here:

  1. Part 1 - Setup and initial trial
  2. Part 2 - Data Collection
  3. Part 3 - Feature Engineering and Selection
  4. Part 4 - Hyperparameter Tuning, Application and API

Prerequisites

Usage

Creating you own models

  1. Change the items_to_predict array in the main() function to the items you wish to use.
  2. Then, run:
python models.py
  1. You should see the .h5 model file created in the models folder along with features.txt file in the models/features folder

Applying the created models

  1. Make sure you have the latest data stored in data/rsbuddy or change the path of DATA_FOLDER in line 101 of application.py
  2. Change the items_to_predict array in the main() function to match the models you created/have.
  3. Then, run:
python application.py
  1. You should see a .csv file created (or have data appended to) in the name of that item in data/predictions.

Running the flask app

  1. Change items in items_predicted array in index() to match the items that you've predicted on
  2. Run:
python flask-app.py
  1. Go to localhost:80 and see your results!

Running the jupyter notebooks

  1. Move the preferred notebook out of the Notebooks foler to the main directory
  2. Run the following command:
jupyter notebook

Scraping your own data

If you wish to scrape your own data the way I've been doing it, run the following script every 2 minutes (for osbuddy):

python osbuddy-ge-scraper.py

OR every 30 minutes (for rsbuddy):

python rsbuddy-ge-scraper.py

You can do this automatically by using crontab if you're on a Linux machine or windows scheduler if you're on a Windows machine.

Credits

Contributions

Community