https://github.com/altescy/automlcli
Simple AutoML command line tool for tabular data
https://github.com/altescy/automlcli
automl cli mlflow python
Last synced: 3 months ago
JSON representation
Simple AutoML command line tool for tabular data
- Host: GitHub
- URL: https://github.com/altescy/automlcli
- Owner: altescy
- License: mit
- Created: 2021-02-23T18:32:44.000Z (over 4 years ago)
- Default Branch: main
- Last Pushed: 2021-05-23T12:19:55.000Z (over 4 years ago)
- Last Synced: 2025-01-26T11:42:19.018Z (8 months ago)
- Topics: automl, cli, mlflow, python
- Language: Python
- Homepage:
- Size: 325 KB
- Stars: 4
- Watchers: 3
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
AutoMLCLI
=========
[](https://github.com/altescy/automlcli/actions?query=workflow%3ACI)
[](https://github.com/altescy/automlcli/blob/master/LICENSE)
[](https://github.com/altescy/automlcli/)
AutoMLCLI is a simple AutoML command line tool for tabular data.
### Features
- Enables you to train and evaluate model / make prediction from command line without writing python scripts
- Read / write data from web or cloud strages
- Mange experimental results with [MLflow](https://github.com/mlflow/mlflow)
- Highly extensible by plugin system
### Supported models
- [TPOT](https://github.com/EpistasisLab/tpot)
- [FLAML](https://github.com/microsoft/FLAML)
## Installation
```
$ pip install git+https://github.com/altescy/automlcli.git
$ pip install "automlcli[all]"
```
## Usage
#### Write config file
```yaml
# config.yml
model:
type: flaml
time_budget: 100
metric: accuracy
task: classification
target_column: target
```
#### Train a model
```
$ automl train \
config.yml \
https://raw.githubusercontent.com/altescy/automlcli/main/tests/fixtures/data/train.csv \
--serialization-dir out
$ ls out
best.json flaml.log metrics.json model.pkl
```
#### Evaluate the trained model
```
$ automl evaluate \
out/model.pkl \
https://raw.githubusercontent.com/altescy/automlcli/main/tests/fixtures/data/dev.csv \
--cv 5 --scoring accuracy --scoring f1_macro
```
#### Make prediction
```
$ automl predict \
out/model.pkl \
https://raw.githubusercontent.com/altescy/automlcli/main/tests/fixtures/data/test.csv \
--output-file predictions.csv
```