Glycoheamoglobin value is actually the most influential feature for the diagonosis of diabetes here.

I shared how to convert the values to integers to show how that works. The original dataset already has an encoded outcome column(which was derived from the established ranges of glycoheamoglobin values.

I get your thoughts about BMI and waist circumference(you can see the correlations matrix)but the interesting thing is that there are people with normal BMI but high waist cm and vice versa.

Thanks for reading through and giving the feedback.