redcican / final_project

EASYML

Auto Machine Learning System

1. Introduce EasyML

EasyML is an Auto Machine Learning web based that help you build Model without any coding.

All you need todo is upload your data, design your experiment like select target, features, score and submit. EasyML will help you generate more than 10 models with evaluation score.

Check bellow video to learn more detail about EasyML system:

There are alot of features need to be added, so it would be great if you can join with me

2. How to use EasyML

Main Page

2.1 Upload Data and analysis your data

After build the model, you need upload data by click on File Management button. System now only accept csv file. I will upgrade soon that help customer can upload other file like excel, text.

To EDA your data, click Open EDA. I used PandasProfiling to help me automatic this report.

2.2 Create Experiment

Click Experiment, select your data using to build model, choose target, features, train and test split ratio and submit.

3. Experiment Detail and Prediction

After Experiment success, you can check the performance at this page. Although, you can make a prediction with best model score.

4. Model Evaluations

By click on models in Leaderboard, you can check more detail about model performance.

This function still implementing. I will release it as soon as possible.

5. How to install

pip install -r src/requirements.txt

Install Redis

Install PostgresSQL

Go to src/mlplatform/settings.py and change database config

Change the database config to your postgresql config

DATABASES = {
    'default': {
        'ENGINE': 'django.db.backends.postgresql',
        'NAME': 'mlplatform',
        'USER': 'postgres',
        'PASSWORD': 1228,
        'HOST': 'localhost',
        'PORT': 5432,
    }
}

Change the Celery config

CELERY_BROKER_URL = 'redis://localhost:6379'
CELERY_RESULT_BACKEND = 'redis://localhost:6379'

6. How to run

Django Database migration

python manage.py makemigrations main
python manage.py migrate main
python manage.py makemigrations optirodig
python manage.py migrate optirodig

Manually add data from script

run

python src/scripts/load_scrap.py

run it three times to add data from csv files.

if __name__ == '__main__':
    #run()
    #run_giesserei()
    run_chemi()

Check PostgresSQL from interface (pg Admin4)

if there are not data available, run

python manage.py migrate --run-syncdb

to sync the dataset

Start django service

python manage.py runserver --noreload

Start worker (new terminal)

celery -A mlplatform worker -s /tmp/tmp.db --pool=solo -l info

Start celery flower service (optional, new terminal)

celery -A mlplatform flower --port=5555