Prediction-Model

Cervical cancer screening prediction using Python

Built a classifier for cervical cancer screening prediction using the Kaggle dataset (https://www.kaggle.com/datasets/loveall/cervical-cancer-risk-classification).

Performed BINARY CLASSIFICATION for each of the 4 target variables individually namely Hinselmann, Schiller, Cytology and Biopsy using two classifiers: SVM and KNN.

Steps Performed:

Data Preprocessing

• Dealt with missing values
• Identified and removed the outliers
• Normalized the data

Data Balancing

• Used SMOTE to balance the classes

Feature Extraction

• Identified useful features and eliminated redundant features
• Reduced the dimensionality of data
• Used PCA to extract the principal components

Classifier

• Used two classifiers for this task: SVM and KNN
• Tested the results by tuning the hyperparameters for each classifier, e.g., regularizer weight for soft-margin in SVM and the value of k in KNN

Evaluation

• Three evaluation metrics are used for each classifier: Accuracy, Precision and Recall
• Confusion matrix for each target variable is plotted

Data Visualization

• Visualized the normalized data distribution using boxplot
• Identified correlated features using correlation heatmap
• Plotted the confusion matrix
• Used seaborn and matplotlib libraries for visualization.