victorperezc/eventml

Pipeline Engine for Event Driven Machine Learning to crawl, preprocess and train ML models.

Event Driven Pipeline Engine

EventML is a modular software to perform Event Driven Machine Learning. It provides the building blocks and engines allowing the user to ensemble pipelines and run them conviniently.

Run locally

To run EventML you will need to install dependecies from the requirements.txt file. You can do so by directly installing them onto your system path or creating a virtual environment.

pip3 install -r requirements.txt

Use Case

We provide with a use case to predict Microsoft Stock Prices with structured event embbedings using Open Information Extraction. You can run this by running the main.py file as

python3 main.py

Dependencies

• pandas and numpy: To handle data frames operations
• OpenIE : Open IE to extract structured events from text
• snscrape : To scrape information from the web

File structure

+-- src
|   +-- pipes
|   |   +-- crawler.py 
|   |   +-- encoders.py
|   |   +-- ml.py
|   |   +-- openie.py
|   |   +-- pipe.py
|   |   +-- preprocessing.py
|   |   +-- sentiment.py
|   |   +-- utils.py
|   |   +-- visualization.py
|   +-- colors.py
|   +-- engine.py
+-- README.md
+-- main.py
+-- requirements.txt