To the list of courses || RASA2018 || To the theme || Estonian

Exercise 3983. Points 4, theme: Relation Between Point Patterns

Open exercise
Traffic accidents with pedestrians are frequent near schools. The attached file contains coordinates of rural schools (Maakoolid.txt), locations of traffic accidents with pedestrians in 2000-2010, and raster layer of state owned roads.

1. Up to which distance are the traffic accidents with pedestrians significantly (p < 0.01) more frequent compared to random locations at the following conditions?
  • If to generate random locations to roads.
  • If to generate random locations in Estonian borders.
  • If to generate random locations to the rectangle which bounds Estonia.
2. In the last case, what affects the frequency of random locations at greater distances (>200 km) from the rural schools?


The neighbours density [O(r) statistic] function enables to calculate distance dependent neigbours density function and to compare it with the expected values in case of random location of the same number of neighbours.
  • Let assume that the given data represent pedestrian accidents on state roads.
  • The mask of Estonian territory is built in to the calculator, raster layer of state roads is in the attached file (Riigiteed_1-0.rst).
  • If a suitability mask is uploaded it must be in Idrisi rst file format, minimum and maximum coordinates matching the study area. Pixel value 1 should present normal or average suitability, 0 values unsuitable area, negative suitability values express deterring locations.
  • The default settings might not fit for a particular task. Experiment with different number and extent of intervals but start from a small number (e.g. 10) of wider (10000m) intervals and small number of iterations (20). For final results, the number iterations should be at least 200.

Each time when a relative indicator is used, the reference variable (denominator) must be clearly understandable. If percentages are mentioned in a text, it must always be clear, what is 100%. For density values, the area unit must be clear.

The confidence interval is not symmetrically along the globally expected value because random locations are generated only to suitable parts of the study area, which proportion decreases at larger distances from abservations.

Log in to send your results and to see the expected answer and responses from other students.