Project summary: This group presents the challenges and rewards of introducing alternative data - satellite images and social networks - for creditworthiness assessment in agriculture.

Project summary: Ascend Interact is a deep-learning based, no-code conversational analytics assistant and insight engine, an Experian first in the rapidly growing “no-code/low-code” market. Built to deliver information orientation for user decision-making, it democratizes analytics/data insights for business/nontechnical users by removing the complex technical barriers and time-consuming analytics workflows needed to query, analyze and present data via direct voice/text interaction without any coding.

Project summary: In this innovative project, we built a model that maximizes the power of consumer information to solve their credit problems. The model transforms the credit information as a sequence of events, similar to a DNA strand, and transforms the features and events into images. With this approach, we bring to models a large amount of information that in traditional models would be lost due to data summarization. This allows us to build a more complete model with all events of consumer information coded within pixels and attributed to event logic and sequencing. Our model uses two methodologies: GeneMap (transforming an individual’s series of market events into something similar to a genetic sequence) and DeepInsight (transforming the genetic sequences into images) and thus, we can predict the default rate using these images inputted into the model, built with deep learning (using convolution neural network). The results are promising.

1. Address Verification: This module uses alternative data (e.g. telecom tower data) to waive any physical checks or document uploads. This is live and generating revenue.
   
   
2. Detect Plus: This module uses telecom call network data to predict application fraud and enables the formation of a closed user group.

Project summary: The CrossCore Signal Hub is the first holistic fraud and identity suite, combining multiple Experian data assets with advanced analytics to keep pace with the rapidly evolving threat of online fraud.

Project summary: Experian Health and DataLabs worked together to build a readmission risk (the risk of returning to hospital within 30 days) model using Health’s claims data. The team built a highly predictive model using new aggregation methods that ultimately will improve care management and reduce avoidable readmissions due to social determinants of health.

Project summary: The project demonstrates how Serasa Experian is able to anticipate market demand and overcome the challenge of not having sole control of how the data is to be processed, building Open Banking products, increasing Serasa’s strength in the Brazilian market and, most important, democratizing credit by using the company’s power advantage and the wide reach of our Serasa Consumer platform.