check_unsystematic_cp: Check for Unsystematic Patterns in Cross-Price Data

In brentkaplan/beezdemand: Behavioral Economic Easy Demand

Check for Unsystematic Patterns in Cross-Price Data

Description

Analyzes data patterns to determine whether there's a systematic relationship between two variables (typically price and consumption), and checks for significant deviations or "bounces" that might indicate unsystematic patterns. This function is particularly useful for analyzing demand elasticity and cross-commodity relationships in behavioral economics.

Usage

check_unsystematic_cp(
  data,
  delta_threshold = 0.025,
  bounce_down_threshold = 0.1,
  bounce_up_threshold = 0.1,
  bounce_none_threshold = 0.2,
  verbose = FALSE
)

Arguments

data	A data frame with columns 'x' and 'y', where 'x' typically represents price and 'y' typically represents consumption or demand.
delta_threshold	Numeric value (default 0.025) that determines what qualifies as a significant trend in the data.
bounce_down_threshold	Numeric value (default 0.1) that determines what proportion of downward deviations qualifies as significant for upward trends.
bounce_up_threshold	Numeric value (default 0.1) that determines what proportion of upward deviations qualifies as significant for downward trends.
bounce_none_threshold	Numeric value (default 0.2) that determines what proportion of deviations qualifies as significant when there's no clear trend.
verbose	Logical value (default FALSE) indicating whether to print intermediate calculations.

Details

The function first determines if the data shows a significant upward trend, downward trend, or no trend by calculating slopes in log-scale. It then checks for "bounces" or significant deviations based on the identified trend type:

- For downward trends: looks for significant upward jumps - For upward trends: looks for significant downward drops - For no trends: looks for deviations both above and below the average

The function uses relative changes (percentage changes) rather than absolute changes to detect bounces, making it suitable for data at different scales.

Value

A data frame with the following columns:

delta_down

Logical. TRUE if there's a significant downward trend.

delta_up

Logical. TRUE if there's a significant upward trend.

delta_none

Logical. TRUE if there's no significant trend.

bounce_up

Logical. TRUE if there are significant upward bounces in a downward trend. NA if not applicable.

bounce_down

Logical. TRUE if there are significant downward bounces in an upward trend. NA if not applicable.

bounce_none

Logical. TRUE if there are significant bounces in no-trend data. NA if not applicable.

bounce_any

Logical. TRUE if any type of bounce was detected.

bounce_above

Integer. Count of upward deviations meeting the threshold criteria.

bounce_below

Integer. Count of downward deviations meeting the threshold criteria.

Examples

# Create a sequence of x values for examples
x_seq <- 10^(seq(-2, 2, length.out = 10))

# Example 1: Flat pattern with some fluctuations
pattern1 <- data.frame(
  x = x_seq,
  y = c(10, 5, 10, 9, 10, 13, 10, 10, 7, 9)
)
check_unsystematic_cp(pattern1)

# Example 2: Upward trend with bounces
pattern2 <- data.frame(
  x = x_seq,
  y = 10 - c(10, 10, 7, 8, 2, 6, 3, 4, 10, 0) # increasing pattern
)
check_unsystematic_cp(pattern2)

# Example 3: Downward trend
pattern3 <- data.frame(
  x = x_seq,
  y = c(10, 10, 10, 9, 8, 5, 3, 2, 0, 0)
)
check_unsystematic_cp(pattern3)

# Example 4: No clear trend with significant deviations
pattern4 <- data.frame(
  x = x_seq,
  y = c(10, 10, 10, 9, 8, 5, 3, 2, 0, 10)
)
check_unsystematic_cp(pattern4)

# Example 5: Perfectly flat line
pattern8 <- data.frame(
  x = x_seq,
  y = rep(10, 10)
)
check_unsystematic_cp(pattern8)

# Analyzing multiple patterns at once
# First combine patterns into one dataframe with IDs
combined_df <- rbind(
  cbind(id = "pattern1", pattern1),
  cbind(id = "pattern2", pattern2),
  cbind(id = "pattern3", pattern3),
  cbind(id = "pattern4", pattern4),
  cbind(id = "pattern8", pattern8)
)

# Then analyze all patterns and combine results
results_list <- lapply(unique(combined_df$id), function(pattern_id) {
  pattern_data <- combined_df[combined_df$id == pattern_id, c("x", "y")]
  result <- check_unsystematic_cp(pattern_data)
  result$id <- pattern_id
  return(result)
})

# Combine all results
all_results <- do.call(rbind, results_list)
print(all_results[, c(
  "id",
  "delta_direction",
  "bounce_direction",
  "bounce_any"
)])

brentkaplan/beezdemand documentation built on June 11, 2025, 3:50 a.m.