الفهرس | يوجد فقط 14 صفحة متاحة للعرض العام |
المستخلص Diabetes is one of the chronic diseases, also know the silent killer, it is continues with the affected person until death and has a long stage without obvious symptoms for the patient. The late diagnosis of the disease leads to the gradual appearance of complications for the patient, which these complications include defects and failures of many vital organs, for example, eyes, liver and nerves. But if the diagnosis is made at an early stage, many of these complications can be avoided or even prevented in some cases. In this study, machine learning methods are used to classify a patient to having diabetes or not at an early stage via logistic regression (supervised machine learning), k-means clustering (unsupervised machine learning) and its results are compared with the response variable for this study, and also used principal components analysis (unsupervised machine learning) before logistic regression and k-means clustering to reduce dimensions, assuming that it improves the classification ratio. The classification rate when applying logistic regression reached 75%, k-means clustering 77%, but when using principal component analysis before logistic regression, the classification rate increased from 75% to 89%, but when using it before k-means clustering. did not lead to an increase in the classification rate. |