Milestone 2

CUSTOMERS, NEEDS, REQUIREMENTS, NEEDS-REQUIREMENTS MAPPING


Project Plan

  • Collect accurate and widespread election data

  • Develop an accurate prediction algorithm for US election results

  • Train algorithm on election data

  • Implement functionality for user defined election variables

Task Breakdown

  • Website development and deployment - Elizabeth Cone

  • AI Model development - EJ Hannah

  • Software architecture - Aleksandar Dimoski

  • Political Research and data collection - Quentin Jimenez

Design Concepts and Selection

  • The scope of the project includes algorithm implementation, testing the model, and creating a website where the prediction algorithm can be easily accessed by users publicly. It's important that the website host be capable of hosting the algorithm itself, and be easy for users to navigate. It's also important for the algorithm to be quick and accurate, as well as trained off relevant data so that it can provide actionable insights.

  • The team has been selecting which machine learning algorithm will work best for our design plan. From the research we have done, we found it best to combine several algorithms, using NLPs, LSTMs, Logistic regression and a neural network to complete our goal. By combing these 4 methods, we will be able to input and output a large amount of data and complete different types of predictions. It will also increase the accuracy of our predictions, by using a generated weighting system the algorithm will be able to determine which algorithm works best for a certain purpose.


Design Concept Matrix

Design

The main focus of the design during milestone 2 is the algorithm itself. The prediction algorithm will take advantage of several common prediction models to ensure it can make full use of various data sources, including web scraping and existing data sources on the relationships between various demographics, regional and national data, and election outcomes.


Hardware and Software Specifications

  • No hardware required for project

  • Website implementation needed for user access

  • Implementation of user defined variables through website

  • Development of prediction algorithm

  • Collection of data through APIs and web-scraping tools

Test Plan

  • Continue model and algorithm planning and development

  • Finalize prototype by January

  • Begin continuous testing, development and training process

  • Continuous testing process while refining algorithm variables

  • Develop and test user defined variables on website.