pointblank
Data quality assessment and reporting.
Create Agent
library(pointblank)
# Create validation agent
agent <- create_agent(df) %>%
col_vals_gt(vars(age), 0) %>%
col_vals_lt(vars(age), 120) %>%
col_vals_not_null(vars(name)) %>%
col_is_numeric(vars(income)) %>%
interrogate()
# View report
agent
Validation Functions
agent <- create_agent(df) %>%
# Value comparisons
col_vals_gt(vars(x), 0) %>%
col_vals_gte(vars(x), 0) %>%
col_vals_lt(vars(x), 100) %>%
col_vals_lte(vars(x), 100) %>%
col_vals_equal(vars(x), 1) %>%
col_vals_not_equal(vars(x), 0) %>%
col_vals_between(vars(x), 0, 100) %>%
# Set membership
col_vals_in_set(vars(status), c("A", "B", "C")) %>%
col_vals_not_in_set(vars(status), c("X", "Y")) %>%
# Null checks
col_vals_null(vars(x)) %>%
col_vals_not_null(vars(x)) %>%
# Pattern matching
col_vals_regex(vars(email), "^[a-z]+@") %>%
interrogate()
Column Checks
agent <- create_agent(df) %>%
# Type checks
col_is_numeric(vars(age)) %>%
col_is_character(vars(name)) %>%
col_is_date(vars(date)) %>%
col_is_logical(vars(flag)) %>%
# Existence
col_exists(vars(id, name, age)) %>%
interrogate()
Row Checks
agent <- create_agent(df) %>%
# Row count
row_count_match(100) %>%
# Distinct rows
rows_distinct() %>%
# Complete rows
rows_complete() %>%
interrogate()
Actions
# Define actions
al <- action_levels(
warn_at = 0.1, # Warn if >10% fail
stop_at = 0.25, # Stop if >25% fail
notify_at = 0.05
)
agent <- create_agent(df, actions = al) %>%
col_vals_not_null(vars(id)) %>%
interrogate()
Reporting
# Get report
get_agent_report(agent)
# Export report
export_report(agent, filename = "report.html")
# X-list (detailed results)
get_agent_x_list(agent)
Informant
# Create data dictionary
informant <- create_informant(df) %>%
info_tabular(
description = "Customer data"
) %>%
info_columns(
columns = vars(id),
info = "Unique identifier"
) %>%
incorporate()
YAML Workflow
# Export to YAML
yaml_write(agent, filename = "validation.yaml")
# Read and run
agent <- yaml_read_agent("validation.yaml") %>%
interrogate()