Malignancy Detection Using SVM Model

• 

  Prediction on Test Set

• 

  Prediction on Training Set

Inspiration

According to U.S. Breast Cancer Statistics, the number of people suffering from Breast Cancer increase every year. Thus, making a model could help radiologist detect cancerous nodules at a primary stage.

What it does

The model accepts inputs as a number of feature variables and calculates the malignancy probability depending on the features.

How we built it

The first step was to collect the data for training the model. We used University of Wisconsin's Breast Cancer dataset. The data had to be preprocessed (feature scaling, categorical encoding). Since we had to many features in our independent variable, we had to use dimensionalility to reduction to reduce those features. We then import the Kernel SVM class, create an object of the class and then fit the model on the dataset. Once the final model is prepared, we used the model to received predictions on new data points. The predictions were plotted for better understanding of the data.

Challenges we ran into

We keep running into errors during the categorical encoding part (we were using categorical encoder but label encoder did the job).

Accomplishments that we're proud of

The accuracy of the model was quite good and classified most of the prediction correctly.

What we learned

We learned about different steps that's involves data processing, looked at various machine learning models and how they work.

What's next for Malignancy Detection Using SVM Model

The next step would be to incorporate the model in some sort of GUI application (may be using Visual Basics) to calculate and see the results in real time.

Built With

• matplotlib
• python
• scikit-learn