# hR: Toolkit for Data Analytics in Human Resources

#### 2021-01-13

Transform and analyze workforce data in meaningful ways for human resources (HR) analytics. The use of two functions, hierarchy and hierarchyStats, is demonstrated below. Convert standard employee and supervisor relationship data into useful formats for robust analytics, summary statistics, and span of control metrics. Install the package from CRAN by running the install.packages("hR") command.

## workforceHistory data

The examples in this vignette use the sample workforceHistory data set. This data set reflects an artificial organization’s historical workforce/employment data. The sample is reduced to a data.table containing one row per active employee and contractor in order to properly iterate over the current hierarchy structure in the following sections.

data("workforceHistory")

# Reduce to DATE <= today to exclude future-dated records
dt = workforceHistory[DATE<=Sys.Date()]

# Reduce to max DATE and SEQ per person
dt = dt[dt[,.I[which.max(DATE)],by=.(EMPLID)]$V1] dt = dt[dt[,.I[which.max(SEQ)],by=.(EMPLID,DATE)]$V1]

# Only consider workers who are currently active
# This provides a reliable 'headcount' data set that reflects today's active workforce
dt = dt[STATUS=="Active"]

# Exclude the CEO because she does not have a supervisor
CEO = dt[TITLE=="CEO",EMPLID]
dt = dt[EMPLID!=CEO]

# Show the prepared table
# This represents an example, active workforce
print(dt[,.(EMPLID,NAME,TITLE,SUPVID)])
#>    EMPLID    NAME     TITLE SUPVID
#> 1: 131356  George   Analyst 199827
#> 2: 199827   Pablo  Director 111355
#> 3: 534441 Rebekah   Analyst 199827
#> 4: 199901 Enrique Associate 199827
#> 5: 268831 Hillary    Intern 131356

## hierarchy

The hierarchy convenience function transforms a standard set of unique employee and supervisor identifiers (employee IDs, email addresses, etc.) into an elongated or wide format that can be used to aggregate employee data by a particular line of leadership (i.e. include everyone who rolls up to Susan).

When format = "long", the function returns a long data.table consisting of one row per employee for every supervisor above them, up to the top of the tree.

hLong = hierarchy(dt$EMPLID,dt$SUPVID,format="long")
print(hLong)
#>     Employee Level Supervisor
#>  1:   131356     1     111355
#>  2:   131356     2     199827
#>  3:   199827     1     111355
#>  4:   199901     1     111355
#>  5:   199901     2     199827
#>  6:   268831     1     111355
#>  7:   268831     2     199827
#>  8:   268831     3     131356
#>  9:   534441     1     111355
#> 10:   534441     2     199827

# Who reports up through Susan? (direct and indirect reports)
print(hLong[Supervisor==CEO])
#>    Employee Level Supervisor
#> 1:   131356     1     111355
#> 2:   199827     1     111355
#> 3:   199901     1     111355
#> 4:   268831     1     111355
#> 5:   534441     1     111355

When format = "wide", the function returns a wide data.table with a column for every level in the hierarchy, starting from the top of the tree (i.e. “Supv1” is the top person in the hierarchy).

hWide = hierarchy(dt$EMPLID,dt$SUPVID,format="wide")
print(hWide)
#>    Employee  Supv1  Supv2  Supv3
#> 1:   199827 111355   <NA>   <NA>
#> 2:   131356 111355 199827   <NA>
#> 3:   534441 111355 199827   <NA>
#> 4:   199901 111355 199827   <NA>
#> 5:   268831 111355 199827 131356

# Who reports up through Pablo? (direct and indirect reports)
print(hWide[Supv2==199827])
#>    Employee  Supv1  Supv2  Supv3
#> 1:   131356 111355 199827   <NA>
#> 2:   534441 111355 199827   <NA>
#> 3:   199901 111355 199827   <NA>
#> 4:   268831 111355 199827 131356

## hierarchyStats

The hierarchyStats function computes summary statistics and span of control metrics from a standard set of unique employee and supervisor identifiers (employee IDs, email addresses, etc.). The resulting metrics and table are accessible from a list object.

hStats = hierarchyStats(dt$EMPLID,dt$SUPVID)

# Total Levels:
print(hStats$levelsCount$value)
#> [1] 4

# Total Individual Contributors:
print(hStats$individualContributorsCount$value)
#> [1] 3

# Total People Managers:
print(hStats$peopleManagersCount$value)
#> [1] 3

# Median Direct Reports:
print(hStats$medianDirectReports$value)
#> [1] 1

# Median Span of Control (Direct and Indirect Reports):
print(hStats$medianSpanOfControl$value)
#> [1] 4

# Span of Control Table
print(hStats\$spanOfControlTable)
#>    Employee directReports spanOfControl
#> 1:   111355             1             5
#> 2:   131356             1             1
#> 3:   199827             3             4