Improving Menstrual Cycle Prediction Accuracy using Advanced Machine Learning Model Methods

Abstract

A menstrual cycle prediction system was developed in this paper. To train the dataset obtained from Kaggle, twenty-two machine-learning algorithms were used. Twelve features were used to obtain the predicting variable. The model accuracy and performance were obtained using six standard evaluation metrics as shown in the result section. 0.9864 was obtained as the R2 score for the Decision Tree algorithm, which was the best amongst other models. This indicates that with the Decision Tree, the ovulation day of women can be predicted with higher accuracy. After saving the model with the highest accuracy, it was deployed on a web application using Python, Django, HTML, CSS. Also, VS Code and Jupyter Noteboook were the IDEs that were used in developing this system.