Neptune + Prophet integration

Experiment tracking for Prophet-trained models.

What will you get with this integration?

• Log, organize, visualize, and compare ML experiments in a single place
• Monitor model training live
• Version and query production-ready models and associated metadata (e.g., datasets)
• Collaborate with the team and across the organization

What will be logged to Neptune?

• parameters,
• forecast data frames,
• residual diagnostic charts,
• other metadata

Resources

• Documentation
• Code example on GitHub
• Example project in the Neptune app
• Run example in Google Colab

Example

Before you start

• Install and set up Neptune.
• Have Prophet installed.

Installation

# On the command line
pip install neptune-prophet

Logging example

# In Python
import pandas as pd
from prophet import Prophet
import neptune
import neptune.integrations.prophet as npt_utils

# Start a run
run = neptune.init_run(project="common/fbprophet-integration", api_token=neptune.ANONYMOUS_API_TOKEN)

# Load dataset and fit model
dataset = pd.read_csv(
    "https://raw.githubusercontent.com/facebook/prophet/main/examples/example_wp_log_peyton_manning.csv"
)
model = Prophet()
model.fit(dataset)

# Log summary metadata (including model, dataset, forecast and charts)
run["prophet_summary"] = npt_utils.create_summary(model=model, df=df, fcst=forecast)

# Stop the run
run.stop()

Support

If you got stuck or simply want to talk to us, here are your options:

• Check our FAQ page.
• You can submit bug reports, feature requests, or contributions directly to the repository.
• Chat! In the Neptune app, click the blue message icon in the bottom-right corner and send a message. A real person will talk to you ASAP (typically very ASAP).
• You can just shoot us an email at [email protected].