Inside PyTorch Forecasting: Deep Learning for FinTech and a better world
Time-series forecasting is fundamental to FinTech—powering use cases like price prediction, risk modelling, fraud detection, and transaction anomaly detection. Yet developers often struggle to build scalable forecasting pipelines without reinventing the wheel.
In this session, I will introduce PyTorch Forecasting (PTF), a widely-used open-source library that simplifies modern forecasting using deep learning. For some context - this library is among the most popular ones in its domain, with more than 4.5k stars on GitHub and 300k downloads a month.
The session aims to provide beginners in this domain with an introduction to the basic principles behind time series forecasting along with a simple hands-on demo of a real world application of PTFs robust pipeline on financial datasets. The session will also provide sneak-peeks into the exciting work behind PyTorch Forecasting's upcoming version 2 API, and some meaningful tips to contributing towards open source development from an active contributor!
Attendees of this talk will gain insights into several aspects.
Understand the basics of time series forecasting and its applications. The talk will dive deep into why time series forecasting is so crucial especially for statistical domains like finance. We will discuss some conventional techniques and some limitations to these techniques.
How PyTorch Forecasting simplifies the process. Attendees will gain understanding on the problems addressed by the library and the numerous features and abstractions that it offers along with a hands-on demo of financial use cases to this library.
Gain insights into the exciting work underway for version 2. We will be discussing design decisions and the thought process behind revamping a library of this scale.
Learn about contributing to open source (in brief) from a fellow developer in this space. Attendees will get actionable insights into contributing towards open source and some tips to get started with this space. Hopefully some of you end up contributing to our library!
Pranav Bhat P
