https://github.com/paulhendricks/gym-inventory

An OpenAI Gym environment for Inventory Control problems
https://github.com/paulhendricks/gym-inventory

openai-gym openai-universe python reinforcement-learning

Last synced: 6 days ago
JSON representation

An OpenAI Gym environment for Inventory Control problems

• Host: GitHub
• URL: https://github.com/paulhendricks/gym-inventory
• Owner: paulhendricks
• Created: 2016-09-20T19:47:19.000Z (about 8 years ago)
• Default Branch: master
• Last Pushed: 2020-03-26T16:35:25.000Z (over 4 years ago)
• Last Synced: 2023-03-01T08:07:02.126Z (over 1 year ago)
• Topics: openai-gym, openai-universe, python, reinforcement-learning
• Language: Python
• Homepage:
• Size: 1.95 KB
• Stars: 40
• Watchers: 2
• Forks: 34
• Open Issues: 4
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# gym-inventory

The [Inventory environment](https://github.com/paulhendricks/gym-inventory) is a single agent domain featuring discrete state and action spaces. Currently, one task is supported:

## Simple Inventory Control with Lost Sales

This environment corresponds to the version of the inventory control with lost sales problem described in Example 1.1 in [Algorithms for Reinforcement Learning by Csaba Szepesvari (2010)](https://sites.ualberta.ca/~szepesva/RLBook.html).

## Future tasks

Future tasks will have more complex environments that take into account:

 * Demand-effecting factors such as trend, seasonality, holidays, weather, etc.

 * Business-specific factors such as lead times, penalties for carrying inventory, etc.

 * Scaling from 1 to millions of SKUs and learning across SKUs.

# Installation

```bash

cd gym-inventory

pip install -e .

```