Electre 1 method on a real example of choosing the best map API

Written by

Suppose that we are building a mapping system to show travel videos from YouTube on a map (see the figure above). To select an interactive system for positioning objects on the map, we will use Electre 1 method for the analysis.
Alternatives that we will consider as part of the analysis for interactive display of objects on the map:
1. Yandex Map – Yandex Map API
2. MapBox – MapBox API
3. Google Map – Google Map API
Criterias:
• Crit.1 – ease of carrying out the procedure of clustering objects.
• Crit.2 — the number of options for setting up the map (layers, options, object options, etc.).
• Crit.3 — cost of requests.
• Crit.4 – ease of customizing the library for the needs of the project in terms of the volume and complexity of the code, accessibility and simplicity of documentation.
Criteria weights
• Weight1 = 15
• Weight2 = 25
• Weight3 = 20
• Weight4 = 40
We will conduct an expert assessment of each of the presented systems for positioning objects on the map according to the four above criteria on a 100-point scale (see Table 1).

Table 1 – Expert evaluation of systems for interactive positioning of objects on the map

Agreement index values
To calculate the values of the agreement indices, the set of criteria I, consisting of N elements, is divided into three subsets:
I+ – criteria where Ai is preferable to Aj,
I= – criteria, where Ai is equivalent to Aj, and
I- – criteria, where Aj is preferable to Ai.
Then, the agreement indices are calculated by dividing the sum of the weights of the criteria from the subsets I+ and I= by the total sum of the weights of the criteria according to the formula (see Fig. 1):

Figure 1 – formula of agreement indices

Table of agreement index values (see Table 2).

Table 2 — Values of agreement indices for systems of interactive positioning of objects on the map

Disagreement indices can be calculated according to the formula (see Fig. 2):

Figure 2 – Formula of disagreement indices

liA – evaluation of alternative Ai by criterion i
Li is the length of the scale according to criterion i.

Table of values of disagreement indices (see Table 3).

Table 3 — Values of disagreement indices for systems of interactive positioning of objects on the map

Let us compose the Kernel of Analysis and carry out the first iteration (see Tables 4 and 5).
Threshold coefficient for the agreement index: С1 = 0.8.
Threshold coefficient for the disagreement index: d1 = 0.08.

Table 4 — Values of agreement indices for the first iteration of the analysis kernel for systems of interactive positioning of objects on the map
Table 5 — Values of disagreement indices for the first iteration of the analysis kernel for systems of interactive positioning of objects on the map

Because for Yandex Map and Google Map relative to MapBox, the agreement indices are greater than or equal to the threshold value C1 (0.8) and at the same time the disagreement indices in the same cells are below the threshold value d1 (0.08), then it is obvious that Google Map and Yandex Map dominate the MapBox alternative, MapBox we exclude the kernel from the analysis and the analysis of the second iteration is already carried out without MapBox.

Second iteration (see tables 6 and 7).

Threshold coefficient for the agreement index: С1 = 0.6.
Threshold coefficient for the disagreement index: d1 = 0.24.

Table 6 — Values of agreement indices for the second iteration of the analysis kernel for systems of interactive positioning of objects on the map
Table 7 — Values of disagreement indices for the second iteration of the analysis kernel for systems of interactive positioning of objects on the map

On the second iteration, it can be seen that Yandex Map and Google Map API are almost on par, especially with regard to the disagreement index – the difference is very small (0.24 versus 0.25), but still there is in favor of Google Map. The higher agreement index of Google Map compared to Yandex Map, which is above the threshold value of 0.6, allows us to say that using Google Map will be the best strategy when developing an intelligent system for interactive positioning of video clips on the map. Google Map is becoming the best alternative.
The general order of the geo positioning API systems in accordance with the study:

Google Map > Yandex Map > Map Box

According to the analysis of alternatives by the Electre 1 method, the best alternative according to the given criteria will be the third alternative – an approach based on the Google Map API, Yandex Map – in second place in preference. In other words, an interactive system for positioning objects on a map that would satisfy the conditions of optimality and development efficiency is the Google Maps API.
Google Maps API was chosen to display videos on the map of the developed intelligent system vOtpuskSam.ru.

Leave a Reply

Your email address will not be published. Required fields are marked *