PyData Toronto February 2024 Meetup

Hey PyData Community,

We are excited to announce our second meetup of 2024 will be on the 27 February 2024 at Northeastern's Toronto campus. Come join us IRL to meet new people and listen to great talks!

Talks:

Akash Shetty / Text Processing on Data for interacting with LLMs:

Cover user journey on how to process data for LLMs and from LLMs using the libraries LangChain and LLaMa Index. I will be going cover concepts of different text extraction process. Levels Of Text Splitting

1. Character Splitting - Simple static character chunks of data
2. Recursive Character Text Splitting - Recursive chunking based on a list of separators
3. Document Specific Splitting - Various chunking methods for different document types (PDF, Python, Markdown)
4. Semantic Splitting - Embedding walk based chunking
5. Agentic Splitting - Experimental method of splitting text with an agent-like system.
6. Alternative Representation Chunking + Indexing - Experimental method of splitting text using derivative representations of your raw text that will aid in retrieval and indexing

---

Shreya Prasad / Causal Inference: Creating counterfactuals and their applications in Paid Marketing

"Correlation is not causation" - is a fundamental axiom we have all heard since our first course in STEM. But how do we quantify causation? The answer lies in constructing counterfactuals, which enable us to observe different potential outcomes for the same individual under a treatment effect. This talk will explore methodologies for measuring causal impact of interventions, with a focus on their application in paid marketing. We'll delve into counterfactuals, confounding factors, and population matching, techniques that allow us to estimate the outcome that would have occurred under different treatment conditions. We'll discuss how to construct these alternative realities using data science methodologies and we'll explore the application of causal inference in geo experiments within the context of paid advertising.