Package {sensorLeak}

Check for sensors shared between training and testing data

Description

Identifies environmental sensors that occur in both training and testing observations. Shared sensors may represent a validation risk when the intended model should generalize to unseen sensors.

Usage

check_sensor_leakage(data, sensor, split)

Arguments

Value

A data frame containing sensors present in both splits, together with their observation counts.

Examples

data <- data.frame(
  sensor_id = c("S1", "S2", "S1", "S3"),
  split = c("train", "train", "test", "test")
)

check_sensor_leakage(
  data,
  sensor = "sensor_id",
  split = "split"
)

Check for spatial proximity between training and testing data

Description

Identifies training and testing observations that are within a specified geographic distance of each other. Close observations may represent a validation risk when a model is intended to generalize to geographically independent locations.

Usage

check_spatial_leakage(data, lat, lon, split, threshold = 1)

Arguments

Value

A data frame containing pairs of spatially close training and testing observations.

Examples

data <- data.frame(
  latitude = c(13.0827, 13.0830, 13.2000, 13.3000),
  longitude = c(80.2707, 80.2710, 80.3000, 80.4000),
  split = c("train", "test", "train", "test")
)

check_spatial_leakage(
  data,
  lat = "latitude",
  lon = "longitude",
  split = "split",
  threshold = 1
)

Check for temporal leakage

Description

Identifies test observations that occur before the end of the training period. Such observations may indicate an invalid temporal validation setup.

Usage

check_temporal_leakage(data, time, split)

Arguments

Value

A data frame containing test observations that occur before the end of the training period.

Examples

data <- data.frame(
  timestamp = as.POSIXct(
    c(
      "2026-01-01 10:00:00",
      "2026-01-01 12:00:00",
      "2026-01-01 11:00:00",
      "2026-01-01 15:00:00"
    )
  ),
  split = c("train", "train", "test", "test")
)

check_temporal_leakage(
  data,
  time = "timestamp",
  split = "split"
)

Check for overlapping temporal windows

Description

Detects temporal overlap between training and testing windows within environmental sensor data.

Usage

check_window_leakage(data, start, end, split, sensor)

Arguments

Value

A data frame containing detected overlapping windows.

Examples

data <- data.frame(
  sensor_id = c("S1", "S1"),
  window_start = as.POSIXct(
    c("2026-01-01 10:00:00", "2026-01-01 13:00:00")
  ),
  window_end = as.POSIXct(
    c("2026-01-01 14:00:00", "2026-01-01 18:00:00")
  ),
  split = c("train", "test")
)

check_window_leakage(
  data,
  start = "window_start",
  end = "window_end",
  split = "split",
  sensor = "sensor_id"
)

Print a sensor leakage audit

Description

Prints a concise summary of the potential validation risks identified by a sensor_audit object.

Usage

## S3 method for class 'sensor_audit'
print(x, ...)

Arguments

Value

Invisibly returns the audit object.

Print a sensor audit summary

Description

Print a sensor audit summary

Usage

## S3 method for class 'sensor_audit_summary'
print(x, ...)

Arguments

Value

Invisibly returns the summary object.

Audit environmental sensor data for leakage risks

Description

Runs multiple leakage checks on environmental sensor data and combines their findings into a single audit result.

Usage

sensor_audit(data, time, sensor, split, lat, lon, spatial_threshold = 1)

Arguments

Value

An object of class sensor_audit containing the results of the leakage checks.

Examples

data <- data.frame(
  timestamp = as.POSIXct(c(
    "2026-01-01 10:00:00",
    "2026-01-01 12:00:00",
    "2026-01-01 14:00:00",
    "2026-01-01 16:00:00"
  )),
  sensor_id = c("S1", "S1", "S2", "S2"),
  latitude = c(13.08, 13.08, 13.20, 13.30),
  longitude = c(80.27, 80.27, 80.30, 80.40),
  split = c("train", "train", "test", "test")
)

audit <- sensor_audit(
  data,
  time = "timestamp",
  sensor = "sensor_id",
  split = "split",
  lat = "latitude",
  lon = "longitude"
)

Example Environmental Sensor Dataset

Description

A small synthetic dataset representing environmental sensor observations with timestamps, sensor identifiers, geographic coordinates, PM2.5 measurements, and train/test assignments.

Usage

sensor_data

Format

A data frame with 12 observations and 6 variables:

timestamp

Timestamp of the sensor observation.

sensor_id

Identifier of the environmental sensor.

latitude

Latitude of the sensor location in decimal degrees.

longitude

Longitude of the sensor location in decimal degrees.

PM25

PM2.5 concentration measurement.

split

Training or testing assignment.

Source

Synthetic example data created for the sensorLeak package.

Examples

data(sensor_data)
head(sensor_data)

Summarize a sensor leakage audit

Description

Provides detailed summary information for a sensor_audit object.

Usage

## S3 method for class 'sensor_audit'
summary(object, ...)

Arguments

Value

An object of class sensor_audit_summary.