McMasterAI / TrafficTracker

A street traffic counter that provides efficient and ethical modelling of how people engage and interact within a given public space.
MIT License
7 stars 2 forks source link

Investigate MAADS Python Library for Kafka Connector #48

Closed Jzar closed 3 years ago

Jzar commented 3 years ago

Overview

As a developer looking to implement Apache Kafka as the data streaming tool in the project, I would like to use the python library utility that is the most useful and feature rich.

Typically, libraries for Kafka connections are Python-Kafka or Confluent-Kafka. MAADS is an alternative that boasts other data science features: https://pypi.org/project/maads/

Automatically analyse your data and perform feature selection to determine which variables are more important than others. Automatically model your data for seasonality Winter, Shoulder, and Summer seasons. Automatically clean your data for outliers. Automatically make predictions using the BEST algorithm (out of hundreds of advanced algorithms) that best model your data. Connect to Apache KAFKA brokers (integrated with MAADS and HPDE) to create topics, produce data to topics, consume data, activate/deactivate topics, create consumer groups. list all brokers statistics, and more.. Automatically optimize the optimal algorithms by MINIMIZING or MAXIMIZING them to find the GLOBAL OPTIMAL VALUES of the independent variables using nonlinear optimization with constraints Perform Natural Language Processing (NLP) on large amounts of text data - and get MAADS to summarize the text or apply deep learning for predictive outcomes. For example, you can tell it to scrape a website, read a PDF, or text data and it will return a concise summary. This summary can be used to refine your modeling and give users an integrated view of their business from a TEXT and ADVANCED ANALYTIC perspective. Or, apply machine learning to text data for deeper insights - such as analysing help desk tickets and uncovering issues before they occur. Or, apply deep learning to security logs and uncover more anomalies or threats in your networks. Do all this in minutes.

Purpose:

If we could use a kafka connector library that was able to simultaneously do a lot of our data transformation and our machine learning, this would be optimal.

Investigation into its usefulness as a kafka connector will be done first. If there are losses in data performance between this library and others, our primary concern would be sending the video data.

Findings will be presented at next scrum.