https://github.com/dkatz23238/randomforestadaptiveexperim
Using Facebook Adaptive Experimentation platform to tune random forest regressors using docker
https://github.com/dkatz23238/randomforestadaptiveexperim
adaptive-learning docker docker-compose hyper-parameter-tuning machine-learning pytorch
Last synced: 2 months ago
JSON representation
Using Facebook Adaptive Experimentation platform to tune random forest regressors using docker
- Host: GitHub
- URL: https://github.com/dkatz23238/randomforestadaptiveexperim
- Owner: dkatz23238
- Created: 2019-05-07T22:13:21.000Z (about 6 years ago)
- Default Branch: master
- Last Pushed: 2019-05-10T23:50:44.000Z (about 6 years ago)
- Last Synced: 2025-01-19T08:28:10.136Z (4 months ago)
- Topics: adaptive-learning, docker, docker-compose, hyper-parameter-tuning, machine-learning, pytorch
- Language: Python
- Size: 4.5 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# ax-container
Dependens:
Python 3.7+
This project does the following:
- Loads a machine learning data set
- Instantiates an Adaptive Experimentation service loop
- Trains and tunes an xgboost.XGBRegressor model on a dataset. The dataset provided by default is house pricing data in buenos aires.
Note: the data set must be ready to process by an sklearn or xgboost algorithm.
## Quickstart
Review the enviornment variables and if needed update the dataset.csv with custom data set. Remember to update the TARGET enviornment variable with the name of the target variable column.
The following enviornment variables can be used and modifed in docker-compose.yml:
- N_TRIALS: the amount of trials to run by adaptive experimentation
- DATASET_PATH: the path to the machine learning dataset
- DATASET_TARGET_NAME: the name of the column that contains the target variable
``` docker-compose up```
or without docker:
cd to app
``` python -m pip install -r requirements.txt && python run.py```
The hyper-parameter tunining process will be printed to screen or loged to the docker compose logs.