Prosper Model Factory Goes Self-Serve

Ohio-based Prosper Insights & Analytics has launched an automated version of its Prosper Model Factory, built on Amazon SageMaker and promising to dramatically simplify and accelerate predictive model development.

Prosper has conducted a nationally representative monthly survey of more than 8,000 consumers since 2003, thus building one of the largest continuously fielded consumer intelligence datasets in the United States. Data captured includes spending intentions and purchase plans; economic sentiment and financial confidence; job expectations and workforce outlook; psychographic indicators such as happiness and impulsivity; media consumption and influence; and future purchase behavior and demand expectations.

The Model Factory was first unveiled in 2019 and was itself the result of eight years of research and development. The Factory combines Prosper's proprietary and predictive zero-party consumer data with the AWS SageMaker advanced analytic development platform, allowing users of the latter to aggregate their own first party data with Prosper information in a 'virtual clean room'.

The new solution, pitched at both technical and non-technical users, provides organizations with direct access to Prosper's data, together with an automated, self-service modeling environment. Combining this with machine learning and deployment infrastructure results in a reduction in timescales from 'weeks to hours,' for a range of purposes including audience targeting models, customer scoring systems, forecasting models, and predictive analytics applications.

Users select features and define target variables from a pre-curated dataset, generate and optimize models, deploy them through serverless real-time inference or batch processing, and automatically generate APIs and deployment endpoints. The platform also supports customer file enhancement and scoring workflows, allowing organizations to append behavioral intelligence signals to existing customer, prospect, financial, workforce, and marketing datasets. Potential applications include audience segmentation and lookalike modeling; customer scoring, churn prediction and LTV modeling; macroeconomic and time-series forecasting; and workforce and hiring projections.

Prosper EVP Phil Rist (pictured) comments: 'This platform combines direct access to a unique behavioral intelligence dataset with an automated modeling and deployment workflow that dramatically simplifies predictive analytics. Organizations can now move much more quickly from behavioral data to actionable predictive insights without the traditional complexity associated with model development and deployment.'

The firm is online at www.prosperinsights.com .

All articles 2006-23 written and edited by Mel Crowther and/or Nick Thomas, 2024- by Nick Thomas, unless otherwise stated.