ditat.io framework for deployment of machine learning models.
For further information visit ditat.io
pip install git+https://github.com/ditat-llc/ditat_ml.git --upgrade
Using ssh
pip install git+ssh://[email protected]/ditat-llc/ditat_ml.git --upgrade
The other option is to clone the repo and install it locally.
pip install -e ./ditat_ml
This framework gives you the flexibility to operate with high-level api (Pipeline) and also with its low-level functions, among others.
Train and Test
from ditat_ml import Pipeline
p = Pipeline()
p.load_data(path='dataset.csv')
p.load_X_y(X_columns, y_columns)
p.preprocessing(**mappings ) # Check documentation for more details.
p.scale()
p.model = YOUR_MODEL_CHOICE
p.train()
Deploy
One you are satisfied with your model's performance, you can deploy it.
p.deploy(name='my_model')
Predict
from ditat_ml import Pipeline
p = Pipeline()
predictions = p.predict(path='dataset.csv', model_name='my_model')
You can also use its low-level functions to give you more flexibility.
from ditat_ml import utility_functions