main.py

import os
import yaml
import logging
import shutil

from datetime import datetime 
from sqlalchemy.engine import URL
#from sqlalchemy.engine.url import URL
from sqlalchemy.pool import NullPool

from triage.util.db import create_engine
from triage.component.timechop import Timechop
from triage.component.timechop.plotting import visualize_chops
from triage.component.architect.feature_generators import FeatureGenerator
from triage.experiments import MultiCoreExperiment, SingleThreadedExperiment


# import os
# os.chdir('donors-choose')
now = datetime.now()
logger = logging.getLogger()
logger.setLevel(logging.INFO)
formatter = logging.Formatter('%(name)-30s  %(asctime)s %(levelname)10s %(process)6d  %(filename)-24s  %(lineno)4d: %(message)s', '%d/%m/%Y %I:%M:%S %p')
fh = logging.FileHandler(f'triage_{now}.log', mode='w')
fh.setFormatter(formatter)
logger.addHandler(fh)


# creating db engine
## check if credentials are stored as environment variables or on yaml file.
environment_variable = os.getenv('PGUSER')
if environment_variable is not None: 
    db_url = URL.create(
            'postgresql+psycopg2',
            host=os.getenv('PGHOST'),
            username=os.getenv('PGUSER'),
            database=os.getenv('PGDATABASE'),
            password=os.getenv('PGPASSWORD'),
            port=5432,
        )
## db credentials not setup on environment variables
else: 
    with open("database.yaml", "r") as f:
        config = yaml.safe_load(f)

    db_url = URL.create(
                'postgresql+psycopg2',
                host=config['host'],
                username=config['user'],
                password=config['pass'],
                database=config['db'],
                port=config['port']
            )

db_engine = create_engine(db_url)

# loading config file
config_file = 'donors-choose-config.yaml'
with open(config_file, 'r') as fin:
    config = yaml.safe_load(fin)

# generating temporal config plot
chopper = Timechop(**config['temporal_config'])

# We aren't interested in seeing the entire feature_start_time represented
# in our timechop plot. That would hide the interesting information. So we
# set it to equal label_start_time for the plot.

chopper.feature_start_time = chopper.label_start_time 

visualize_chops(chopper, save_target = 'triage_output/timechop.png')

# creating experiment object
## Multi core
experiment = MultiCoreExperiment(
    config = config,
    db_engine = db_engine,
    project_path = 's3://dsapp-education-migrated/donors-choose',
    n_processes=2,
    n_db_processes=2,
    replace=True,
    save_predictions=False
)
# If single threaded, uncomment the following lines
# experiment = SingleThreadedExperiment(
#     config = config,
#     db_engine = db_engine,
#     project_path = 's3://dsapp-education-migrated/donors-choose',
#     replace=True,
#     save_predictions=False
# )

# Creating the Triage experiment Report
def generate_experiment_report():

    # Path to where you save the notebook template in 
    template_path = './notebooks/experiment_summary_report_template.ipynb'

    # Specify where you will save the executed notebook (recommend not to overwrite the template)
    output_path = f'./notebooks/experiment_summary_report_{now.strftime("%Y%m%d")}.ipynb'

    shutil.copyfile(template_path, output_path)

    os.system(f'jupyter nbconvert --execute --inplace --to notebook {output_path}')
    os.system(f'jupyter nbconvert {output_path} --to html')

experiment.validate()
experiment.run()
generate_experiment_report()