In GIS there are 4 main data classification methods: 1) Natural Breaks, 2) Equal Intervals, 3) Standard Deviation, and 4) Manual Breaks. Natural Breaks is the default classification in GIS, however, it’s not always the most appropriate. When choosing classifications, we have to be careful and think of how to represent data so it’s most ethical and representative of the message we want to deliver with our map. Figure 1 shows all 4 methods used to show Vancouver dwelling costs. Even though the data is identical for all 4, each map represents Vancouver dwelling costs in a different way.
HOW TO CHOOSE A METHOD
We should consider 6 questions when choosing a method:
- Does the method take into consideration the distribution of data?
- Does it make it easier to understand data?
- Does make computations easier?
- Does it make the legend easier to read?
- Is it appropriate for our selected number of classes?
- Is this method the most ethical?
Taking into consideration the 6 questions above, the selected method might be different in different fields. For example, if both a journalist and a real estate agent used the same data of dwelling costs for Vancouver they would probably choose different methods. If I were a journalist, I would choose natural breaks. The data is skewed and the most accurate method for analysis would be natural because natural breaks method generates homogeneitywithin classes and heterogeneity between classes. Frankly, as a real estate agent, I would use manual breaks for 2 reasons: 1) To round up the numbers and make it easier for customers to read and find the best fit for their budget 2) For higher sales and advertisement, I could manipulate data in a way that the agency’s properties fall into the cheaper priced areas. In that way, customers would think they’re buying a reasonably priced property relative to the rest of the city. However, the latter is far from ethical and data classification should not be used to manipulate any decisions. Therefore, a reasonable and ethical manual break method would be best.