Malignant Tumor Machine Learning Model

The goal of this project was to create a machine learning model that would solve a real world problem.  Determining whether or not a tumor is cancerous in nature can be a life-changing diagnoses. If the tumor is mis-diagnosed as benign, the cancer can be left untreated. While the opposite, a tumor being mis-diagnosed as malignant, can still expose the individual to the difficult trials that come from cancer treatment when it's not necessary.

The final product, while not near the true scale or accuracy necessary to truly  be applied to solve the problem, it is a good showcase of what machine learning can accomplish, especially in the medical field. 

While working on this project there were a lot of speed bumps. One of the major ones was finding a way to simply deploy my application without paying for any cloud deployment services like Microsoft Azure or Amazon AWS. While these services are truly amazing and have a lot of features, I found that no longer having a .edu account with them was going to be so costly.

As such, I turned away from the familiar areas and decided to take a leap and learn Jupyter Notebooks. Using the notebooks, I was able to implement my project as well as present it. While the notebooks were difficult to figure out at first, they quickly became a lot easier to work with as I began to understand the intricacies within them. 

Deciding on the Machine Learning model to use for this project was a difficulty as well. With so many great options, like Linear Discriminatory Analysis(LDA),  Convolutional Neural Networks (CNNs), or even Random Forests, it was difficult to reach the decision of using an SVM model. I ended up testing multiple in the IDE version of my project before I finally decided on an SVM because of accuracy and efficiency.