geo_detect: Detect regions that form an unusual spatial aggregation of...
In cpacc/geode: Functions for Mapping and Analysis of Geospatial Data

Description Usage Arguments Examples

View source: R/geo_detect.R

This function identifies clusters in data organized and aggregated into regions. The Kulldorff cluster detection method is used, which creates groupings of regions by consecutively aggregating nearest-neighbour areas until a set proportion of the total population is reached. The number of cases or events in the aggregation, compared to the expected count (as calculated or user-specified) is used to calculate the likelihood based on either a binomial or Poisson model. The output is the group of neighbouring regions that form the most likely cluster and, any secondary clusters. Optionally a map of the clusters can be generated.

geo_detect(
  data,
  counts,
  pop,
  expected_counts = NULL,
  method = "kulldorff_binomial",
  plot = TRUE,
  legend_title = "Clusters",
  plot_title = NA,
  transparency = NA,
  hover_id,
  interactive = TRUE
)

`data`	Name of input data for analysis. REQUIRED.
`counts`	Name of the data column that contains the observed counts that will be tested for clustering. REQUIRED.
`pop`	Name of data column that contains population sizes to be used for each region. For a cluster analysis, all regions must have a population size > 0. REQUIRED
`expected_counts`	Name of the data column that contains the expected counts for each region. Only used when method = "kulldorff_poisson" is specified. If no data column is provided, the expected count for each region is calculated automatically as proportionate share of the total counts based on population size. OPTIONAL.
`method`	Statistical method to be used for cluster detection. Available methods are "kulldorff_binomial" (default) and "kulldorff_poisson". OPTIONAL.
`plot`	Should an output plot be created showing clustered regions? Default is TRUE. If FALSE, output returns a listing of the regions included in the most likely cluster and the top two secondary clusters. OPTIONAL.
`legend_title`	Title of plot legend. If no value is given, 'Cluster' is used as the default title. OPTIONAL.
`plot_title`	Main title of plot. If omitted, no title is shown. Only applies to static image maps. OPTIONAL.
`transparency`	value between 0 and 1 (default) that defines transparency of map colours, from transparent (0) to opaque (1). Intermediate values allow more or less visibility of underlying (reference) layers in interactive maps. OPTIONAL.
`hover_id`	Name of the data column containing the label or value to be shown when hovering over a particular geographic unit. For interactive maps only. Default is the value given in the specified 'counts' column. OPTIONAL.
`interactive`	TRUE (default) or FALSE. Indicating whether map should be generated as an interactive view or as a static image. OPTIONAL.

# Cluster analysis and map of event rate based on Kulldorff binomial model ----------------
geo_detect(data = my_geo_data,
  counts = events_data_column,
  pop = pop_data_column,
  method = "kulldorff_binomial",
  plot = TRUE)

# Cluster analysis and map of event rate based on Kulldorff Poisson model -----------------
geo_detect(data = my_geo_data,
  counts = events_data_column,
  pop = pop_data_column,
  expected_counts = expected_data_column,
  method = "kulldorff_poisson",
  plot = TRUE)

# Cluster analysis and return listing of regions included in top three clusters -----------
geo_detect(data = my_geo_data,
  counts = events_data_column,
  pop = pop_data_column,
  method = "kulldorff_binomial",
  plot = FALSE)