Data monitoring

From missing values to dataset completeness, and population decay. Everything you’ll need to ensure that your models have what they need to make their best predictions.

Data quality

I want to monitor missing values for all entities in all LTV models across VIP customers

With data quality monitors your team can quickly uncover data pipeline issues and detect when features, predictions, or actual data points don’t conform to expectations.

Detect pipeline issues with missing and fixed values.
Detect input changes with new and out-of-range values.

Dataset schema

I want to monitor removed fields for S3 in 3 models across entire set

Easily secure the structure of your training to inference schema across datasets, data connectors, and streaming to make ensure consistent tracking and visibility of your data and machine learning processes.

Catch deleted/updated fields.
Track schema changes.
Monitor growth rates.

Data activity

I want to monitor model stillness for prediction in 3 models across all segments & entire set

Measure the activity levels of your ML models and their operational metrics to catch in real-time variances potentially correlated with model issues and technical bugs.

Ensure data is being updated on schedule.
Identify peaks and drops in volume.
Pinpoint model stillness.

Population changes

I want to monitor growing predictions in churn models across NA & EU accounts

Stay on top of population representation to achieve the best results for all sub-populations your model serves with monitors to identify growing, decaying, and under-performing segments.

Get started

Identify growing populations.
Identify underperforming segments.

Get started

Try the community edition

No credit card required. FREE forever.

1

Sign up for Superwise

Easily get started with a free community edition account.

!pip install superwise

2

Build your project

import superwise as sw
project = sw.project(“Fraud detection”)

model = sw.model(project,“Customer a”)

policy = sw.policy(model,drift_template)

3

Start monitoring

Fraud detection

Entire population drift – high probability of concept drift. Open incident investigation →

Fraud detection

Segment “tablet shoppers” drifting.
Split model and retrain.

FREE community edition

Featured resources

Blog

The real deal: model evaluation vs. model monitoring

Model evaluation and model monitoring are not the same thing. They may sound similar, but they are fundamentally different. Let’s see how.

Data drift detection basics

Drift in machine learning comes in many shapes and sizes. Although concept drift is the most widely discussed, data drift is the most frequent, also known as covariate shift. This post covers the basics of understanding, measuring, and monitoring data drift in ML systems. Data drift occurs when the data your model is running on…

5 ways to prevent data leakage before it spills over to production

Data leakage isn’t new. We’ve all heard about it. And, yes, it’s inevitable. But that’s exactly why we can’t afford to ignore it. If data leakage isn’t prevented early on, it ends up spilling over into production, where it’s not quite so easy to fix. Data leakage in machine learning is what we call it…