Repository for the first homework of the Advanced Python course API can do the following requests:

/classes

returns available classes

/models

returns available models stored in 'models' directory

/create_model

args={name:name,class_name:class_name,params:params}

creates new model with specified arguments, "class name" arg is mandatory, params should be a string dictionary with hyperparameters of the chosen sklearn model class

Example:

params='{"learning_rate": 0.01,...}'

/delete

args={name:name}

deletes model with specified name

/train

args={"name":name},json

training data should be sent in json file and contain 'target' column, example using requests:

requests.get(url,json=data)

Function was tested with pandas dataframe that has been converted to dict using:

dd.to_dict(orient="records")

/predict

args={name:name},json

data should be sent in the same format as in '/train'

/results

args={"task_id":task_id}

returns results or state of the task

Dockerhub https://hub.docker.com/repository/docker/spielmeister/advancedpythonhw2api API https://hub.docker.com/repository/docker/spielmeister/advancedpythonhw2worker Worker

In HW2 added MongoDB for model storage, celery tasks and docker. App can be build with docker-compose, 4 containers are created(Mongo,app,worker,redis)