/spatial_data_reviewer

building spatial data model with R

Primary LanguageR

Spatial Data Analyzer

Hubert Śmieciuch

Maciej Świtała

2020

Repository: https://github.com/superHubert/spatial_data_reviewer
Shiny server app: http://ec2-34-207-189-82.compute-1.amazonaws.com:3838/pov_spatial_model/app/

An app is a tool for quick and hand spatial data overview.

Four tabs of the Spatial Data Reviewer:

  1. Data
  2. Map
  3. Analyses
  4. Model

Tab one - Data.

Here user provides a data file in one of two formats .csv or .rds . It has two main functions:

  1. Getting familiar with data with filtering by year, unit and variable
  2. Checking for missing values per year and variable for given unit(s), which can be helpful in choosing a right variable and year for further analysis

For review one can load data06_18_contig_na_fill.RDS file in data directory. Spatial ID should be selected as: teryt https://github.com/superHubert/spatial_data_reviewer/blob/master/data/data06_18_contig_na_fill.RDS

1

2

Tab two - Map

Here user may provide spatial data files (required/allowed files are: .shp, .prj, .shx, .dbf).

Map tab provided shapefile spatial units are matched by Shp units ID input, with Data tab provided file elements assigned by selected Spatial ID. It is important to select appropriate variables, otherwise map will not be created.

For review one can load powiaty4 shapefiles in data directory. Shp units ID will be by defould sellected appropriately.

Main features are:

  1. Preparing interactive map (based on highcharter package), which can be downloaded in .hmtl format

  2. Preparing static map based on ggplot package, which can be downloaded in .png format.

  3. Maps can be customized in form of palette color, various variable bucketing methods, custom title, sizes of legend in static map, reversed colors of scale.

  4. One can prepare map automatically with default settings after providing data files and selecting year and variable (automatic split in pretty mode)

3

4

5

6

7

8

Tab three - Analyses

Here another step into data is done, by:

  1. Global Moran statistic of selected variable
  2. Histogram of selected variable
  3. Missing values plot (percentage of missing values)
  4. List of mostly correlated variables with selected one
  5. Scatterplot of variable against another one

9

Tab four - Model

A tab to get summary of spatial model of given functional form and family after selection of variable, and year. This tab also stepwise variable selection if performer and returns set of recommended variables, for this reason it requires pointing out non-numeric variables in Except input

Possible are model types are:

  1. ols
  2. manski
  3. sac
  4. sdem
  5. sem
  6. sdm
  7. sar
  8. slx

Functional forms:

  1. Default y ~ x

  2. Multinomial y ~ x + x^2+x^3+x^4

  3. Power log(y) ~ log(x)

  4. Exponential log(y) ~ x

10