- Cancer is a dreadful disease. It is expensive to treat and even after treatment, there is no guarantee for recovery. In most cases, the victim succumbs to death. It is with this background, having witnessed a friend's mum suffering from cancer and host of other people I know, that made me to try use machine learning to help in providing solution for future cases.
What it does
It collects 10 user required features. These are measurement from a human cell. Its radius, perimeter, texture, etc. After filling the form with these values and submit it, the system predicts whether the sample cell(s) was malignant with cancer or benign.
How I built it
User Interface built with html and css. Its basically a form that collects the cell features as inputs to the model for cancer prediction. The backend is built with flask which contains the logic for handling requests from the client. At the backend I added the .pkl file which contains saved trained model. This is helpful so that to avoid retraining the model when new input is submitted. I finally deployed the system to heroku which is a PaaS
Challenges I ran into
Figuring out how to integrate trained model with flask was a difficult at first but I managed to go around it.
Accomplishments that I'm proud of
I can now train a machine learning model, a skill I did not posses before the sprint.
What I learned
About machine learning. Importing data to jupyter notebook, exploring and cleaning data, scaling features, encoding categorical data, creating and training model, and evaluating accuracy and precision of the model.
What's next for breast-cancer-predictor
Smoothing the imbalanced data, adding Authentication system, storing every prediction made and analysis on predictions. Finding out how it can find its way to the market.