Fidelity: AI Based Anomaly Detection

Koshkarova, Aruzhan; Sirois, Mitchell; Kumar, Nathan; Zhou, Austin

Student Work

Fidelity: AI Based Anomaly Detection

Public Deposited

This project’s goal is to create a machine learning model that predicts data anomalies within Fidelity’s assets and flows dataset. Anomalies occur in three models that are utilized in Snowflake: the TA model (customer holding), FA model (Fidelity fund holding), and Look Through model (consolidated holdings). The models reflect day to day changes in account and fund balances. Often, there are discrepancies in the data due to the process of patching the models together. Fidelity employs a team that is tasked with ensuring data integrity, which includes identifying errors in the models. In the present, errors are time consuming to find and fix. The MQP team proposed AI detection model solutions to quickly and accurately identify anomalies in the TA/FA data consolidation process.

This report represents the work of one or more WPI undergraduate students submitted to the faculty as evidence of completion of a degree requirement. WPI routinely publishes these reports on its website without editorial or peer review.

Creator

Subject

Publisher