Cancer Classification

Using the power of Machine learning to train a model, which can accurately classify the tumors as Malignant or Benign.

Model is trained on a dataset which contains around 30 Feature data points grabbed from snapshots of images of Tumor cells, which are collected by a process called as Fine needle Aspirate(FNA Technique). The resulting trained model will have a high degree of accuracy and can be used to accurately predict the class of tumour on the test data set as well as any future genaralized dataset. Click here to visualise the classification.

Tools Used

• Jupiter Notebook - for coding and visualizing the data

Packages used

• Pandas - for Manipulation of data frames
• Numpy - for Statistical analysis
• Matplot - for Data visualization
• Seaburn - for Statistical Data visualization

Source of Training data-set

• Sklearn Library containing 569 different datasets with corresponding 30 different Feature values.
• UCI Machine Learning Repository

Optimization Techniques used to improve the model

• Data Normalization
• SVM Parameter Optimization(C Param + Gamma Param)- using GridSearchCV

Accuracy Obtained

After deploying the above mentioned optimization techniques, an accuracy of 97% was obtained, with a classification error only in the Type - 1 Error plane of the Confusion matrix.