Unit 8
- Final Assignment - Define ML Problem and Build Own Model(s)
For the last assignment as part of the Intro to Machine Learning Course provided by the Break Through Tech program, I was tasked to define my own ML problem utilized course-approved datasets.
Initially, I was inspired by our Neural Network Sentiment Analysis notebook, so I wanted to create a neural network that could predict the author based on a dataset of book reviews. The bookReviews dataset had to be reoganized and cleaned for proper author and book columns. I realized this was time-consuming. I had trouble verifying if a reader was referring to a book series or if they even mentioned a book title. Essentially, I had no idea which authors and which books they were referring to.
I had to pivot, so I changed my plan and defined a new problem: to predict if an individual's income was greater than 50k, given a variety of personal data from the USA census.
In the Unit 8 folder, Lab8_DefineAndSolveMLProblem.ipynb contains my process and analysis, or free feel to view the pdf version under a similar name.
You can clone this repo to run the notebook yourself to see the models + pickle the models to experiment some more (Pickling happens at the very last cell and will be added to the repo).
P.S. If you prefer, the notebook can be run on the web with Github codespaces for free (be wary of going over Github's limit).