Thanks for the strong interest in the details. Just as a background, I’m a data analyst working for the government of another country. I have also lived in five countries and have seen how governments use data in their policy-making, instead of pure guessing or worse, emotions.

To answer your questions:

1. You were looking at the wrong data sets. There are annual data for enrollment figures, which are sliced and diced based on region, division/district, municipality, and eventually schools. The logic is pretty basic, i.e., the grade 1 students of year X should be the grade 2 students of year X+1, the grade 2 students of year X should be the grade 3 students of year X+1, and so on. For this analysis you need data sets from different years to make a comparison, and in this case we compared between 2013 to 2015. The algorithm is straightforward but you need R programming or complex MS Excel processing to arrive at the numbers (I can teach you if there’s an opportunity). This process is pretty standard with how UN or other organisations compute dropout rates.
2. WHO, UN, and other economic organisations use death and disease figures per capita as a standard. Two cities where one out of ten dies due to a certain disease should be equally alarming, regardless of the size of both cities. If you insist however on adding this dimension in the analysis, feel free to do that, as land area data are easily accessible (even wikipedia I think has it). My guess is it will not make so much difference, or even make Davao look worse, as it has one of the more loose density in the country.
3. You are correct, more than the absolute numbers, it’s good to compare these assault/murder deaths per capita. When you actually do that, the numbers of Davao would even look higher and closer to the NCR counterparts, while that of other cities will remain low. I guess the purpose why absolute numbers were shown, to emphasise on “instances” — which would go back to the premise earlier. Regardless of the size or population of a city, as long as X number of people dies or X number of murders happen every day/week/month in any city, the cause of alarm should be comparable. And to compare Canada, an entire country, with 3 murder deaths per week, is of course much better than Davao, just one city, with 4 murder deaths per week. (assuming the figure you gave is correct, I didn’t check its accuracy).