Package 'farr'

Description

A data set containing dates and descriptions for AAERs.

Usage

aaer_dates

Format

A tibble with 3,386 rows and 4 variables:

aaer_num

AAER number

aaer_date

Date

aaer_desc

Description

year

Year of AAER

Details

The SEC index was refreshed through October 8, 2026. Exact duplicate rows are removed; distinct entries sharing an AAER number are retained. Source-backed corrections retain November 16, 1999 for AAER-1209 and restore AAER-1470 dated October 23, 2001. Historical entries missing from the current index are preserved, including three intentionally omitted numbers with missing dates. The malformed historical identifier AAER-x2218 is replaced by AAER-2218. Raw snapshots, historical supplements, and the correction table with release links are maintained in data-raw; see also the refresh investigation. Index dates have not all been checked against the underlying releases.

Description

A data set containing AAER firms-years used in Bao et al. (2020).

Usage

aaer_firm_year

Format

A tibble with 415 rows and 4 variables:

p_aaer

AAER identifier

gkvey

GVKEY (firm identifier)

min_year

First affected year

max_year

Last affected year

Source

doi:10.1111/1475-679X.12292

Description

A data set containing the dates of Apple media events since 2005.

Usage

apple_events

Format

A tibble with 47 rows and 3 variables:

event

Description of event

event_date

First date of event

end_event_date

Last date of event

Source

https://en.wikipedia.org/wiki/List_of_Apple_Inc._media_events

Description

A function returning AUC.

Usage

auc(scores, response)

Arguments

Value

A named numeric vector containing the area under the ROC curve.

Source

https://blog.mbq.me/augh-roc/

https://stackoverflow.com/questions/4903092/calculate-auc-in-r

Examples

scores <- c(0.1, 0.4, 0.35, 0.8)
response <- c(0, 0, 1, 1)
auc(scores, response)

Description

A data set containing fundamental financial information for Australian banks.

Usage

aus_bank_funds

Format

A tibble with 283 rows and 7 variables:

gvkey

GVKEY (firm identifier)

datadate

Fiscal year-end

at

Total assets

ib

Income before extraordinary items

xi

Extraordinary items

do

Income from discontinued operations

Description

A data set containing fundamental financial information for Australian banks.

Usage

aus_bank_rets

Format

A tibble with 3,047 rows and 4 variables:

gvkey

GVKEY (firm identifier)

datadate

Last trading date of month

ret

Stock return for month

mkt_cap

Market capitalization on datadate

Description

A data set containing identifying information for 10 Australian banks.

Usage

aus_banks

Format

A tibble with 10 rows and 3 variables:

gvkey

GVKEY (firm identifier)

ticker

Stock exchange ticker

co_name

Bank name

Description

Firm-years in RDD analysis of Bloomfield (2021).

Usage

bloomfield_2021

Format

A tibble with 1,855 rows and 2 variables:

fyear

Fiscal year

permco

CRSP firm identifier (PERMCO)

Source

doi:10.1111/1475-679X.12346

Description

Counts of Stack Overflow questions by tag and year, covering 2008–2018. A question can have multiple tags, so summing tag counts does not give the number of distinct questions. The annual total is repeated for each tag's row in a year.

Usage

by_tag_year

Format

A tibble with 40,518 rows and 4 variables:

year

Year

tag

Tag

number

Number of questions with tag during year

year_total

Total number of questions across all tags during year

Description

A simulated data set related to camp attendance.

Usage

camp_attendance

Format

A tibble with 1,000 rows and 2 variables:

id

Student identifier

camp

Indicator for student attendance at camp

Description

Data on whistleblowers and enforcement actions from Call et al. (2018).

Usage

cmsw_2018

Format

A tibble with 1,133 rows and 31 variables:

recid

CMSW record identifier

firmpenalty

The total firm civil and criminal monetary penalties assessed against the firm, its parent and subsidiaries consisting of disgorgement, prejudgment interest, civil fines, criminal restitution, and criminal fines in millions of dollars

otherpenalty

The total firm civil and criminal monetary penalties assessed against the agent firms and/or respondents (e.g., the audit firm, bankers, suppliers) in connection with the financial misrepresentation of the target firm, in millions of dollars

emppenalty

The total civil and criminal penalties assessed against all employees consisting of disgorgement, prejudgment interest, civil fines, criminal restitution, and criminal fines in millions of dollars

empprisonmos

Total incarceration consisting of jail, prison, home detention, and halfway house in months imposed upon employee respondents named in the enforcement action

selfdealflag

An indicator variable equal to one if the violation includes self-dealing such as embezzlement and theft by respondents and equal to zero otherwise

blckownpct

The percentage of blockholder ownership, defined as owners with at least five percent of common shares outstanding from the last 10-K or DEF 14A prior to the first public announcement the firm may be (is) subject to a regulatory enforcement action

initabret

The value-weighted market-adjusted return measured at the close of trading on the initial public announcement date that the firm may be (is) subject to a regulatory enforcement action

wbflag

An indicator variable equal to one if a whistleblower is associated with the enforcement action and equal to zero otherwise

tousesox

Post-SOX action flag

lnvioperiod

The natural logarithm of the total time the violation occurred in months as indicated in the regulatory enforcement proceedings

bribeflag

An indicator variable equal to one if the enforcement actions includes charges under the Foreign Corrupt Practices Act for bribery of a foreign official and zero otherwise

mobflag

An indicator variable equal to one if violation or any of the respondents were associated with a known organized crime family and zero otherwise

deter

An indicator variable equal to one if the violation includes an offense for either option backdating, insider trading, or an offense related to an offering, IPO, merger, or reverse merger and equal to zero otherwise

lnempcleveln

The natural logarithm of the total number of C-level respondents (e.g. CEO, COO, CFO, CAO, CMO, and CIO) named in the enforcement action

lnuscodecnt

The natural logarithm of the total number of unique code sections and rules violated (charges) associated with the enforcement action

viofraudflag

n indicator variable equal to one if fraud under 15 USC §§ 77q, 78j(b), or rules promulgated thereunder are included among the charges in the enforcement action

misledflag

An indicator variable equal to one if the violation included violations of 17 CFR 240.13b2-2 that prohibits materially false or misleading statement to an accountant in connection with the preparation of financial statements and zero otherwise

audit8flag

An indicator variable equal to one if the misreporting firm used a Big N auditor, and equal to zero otherwise

exectermflag

An indicator variable equal to one if the firm terminated an executive respondent as a result of the violations and equal to zero otherwise

coopflag

An indicator variable equal to one if the firm received credit in the assessment of penalties for cooperation as stated in regulatory enforcement documents during the course of the investigation and equal to zero otherwise

impedeflag

An indicator variable equal to one if regulators acknowledged they were deliberately misled and/or charges were included for lying to investigators and equal to zero otherwise

pctinddir

The percentage of the firm's directors that are independent from the last 10-K or DEF 14A prior to the first public announcement the firm may be (is) subject to a regulatory enforcement action

recidivist

An indicator variable equal to one if the firm was previously the subject of a securities regulatory enforcement action and equal to zero otherwise

lnmktcap

The natural logarithm of the market value of equity measured in millions of dollars prior to the first public announcement that the firm may be (is) subject to a regulatory enforcement action

mkt2bk

The sum of market value of equity plus total assets minus total debt divided by total assets with market value determined below and total assets and total debt measured at the last fiscal year end prior to the first public announcement the firm may be (is) subject to a regulatory enforcement action

lev

Total debt divided by total assets measured at the last fiscal year end prior to the first public announcement the firm may be (is) subject to a regulatory enforcement action

lndistance

The natural logarithm of the distance in miles from the location of the firm's headquarters to the offices of the regulator assigned to the geographic area of the firm's headquarter location (closer of the SEC Regional Office or DOJ U.S. District Attorney).

ff12

Fama-French industry code (12-industry)

wbsource

Whistleblower data source

wbtype

Whistleblower type: tipster or nontipster

Source

doi:10.1111/1475-679X.12177

Description

A data set containing data about accruals for 2,000 firms.

Usage

comp

Format

A tibble with 16,237 rows and 14 variables:

gvkey

GVKEY (firm identifier)

datadate

Fiscal year-end

fyear

Fiscal year

big_n

Indicator for Big Four auditor

ta

Total accruals (scaled by assets)

roa

Return on assets

cfo

Cash flow from operating activities (scaled by assets)

size

Size

lev

Leverage

mtb

Market-to-book ratio

inv_at

1/Total assets

d_sale

Change in revenue

d_ar

Change in accounts receivable

ppe

Property, plant & equipment (scaled by assets)

Description

A function returning sensitivity and precision.

Usage

confusion_stats(scores, response, predicted = NULL, k = NULL)

Arguments

Value

A named numeric vector containing sensitivity (the fraction of positive responses detected) and precision (the fraction of predictions classified as positive that have a positive response).

Examples

scores <- c(0.1, 0.4, 0.35, 0.8)
response <- c(0, 0, 1, 1)
confusion_stats(scores, response)

Description

Writes the result of a DuckDB query (or table) to a Parquet file using DuckDB's ⁠COPY (SELECT ...) TO ... (FORMAT PARQUET)⁠ and returns a dbplyr table that reads the written Parquet file.

Usage

duckdb_to_parquet(data, name, schema, data_dir = NULL, overwrite = TRUE)

Arguments

Value

A dbplyr table (lazy) reading the written Parquet file.

Examples

if (requireNamespace("duckdb", quietly = TRUE)) {
  local({
    conn <- DBI::dbConnect(duckdb::duckdb(),
                           config = list(threads = 2, home_directory = tempdir()))
    on.exit(DBI::dbDisconnect(conn, shutdown = TRUE), add = TRUE)
    data_dir <- tempfile("farr-example-")
    on.exit(unlink(data_dir, recursive = TRUE), add = TRUE)
    data <- dplyr::copy_to(conn, data.frame(id = 1:3, value = c(10, 20, 30)),
                           "example_data")
    saved <- duckdb_to_parquet(data, "example_data", schema = "",
                               data_dir = data_dir)
    dplyr::collect(saved)
    loaded <- load_parquet(conn, "example_data", data_dir = data_dir)
    dplyr::collect(loaded)
  })
}

Description

A data set containing the GVKEYs and datadates for firm-years used in Fang, Huang and Karpoff (2016).

Usage

fhk_firm_years

Format

A tibble with 60,272 rows × 2 variables.

gvkey

GVKEY (firm identifier)

datadate

Fiscal year-end

Source

doi:10.1111/jofi.12369

Description

A data set containing the tickers, GVKEYs, and treatment indicator for the SHO pilot program.

Usage

fhk_pilot

Format

A tibble with 3,030 rows × 4 variables.

ticker

Ticker

gvkey

GVKEY (firm identifier)

permno

PERMNO (CRSP security identifier)

pilot

SHO pilot program treatment indicator

Source

doi:10.1111/jofi.12369

Description

Calculate deciles for a variable.

Usage

form_deciles(x)

Arguments

Value

An integer vector the same length as x, with decile assignments from 1 (lowest values) to 10 (highest values).

Examples

library(farr)
library(dplyr, warn.conflicts = FALSE)

df <-
    tibble(x = rnorm(100)) %>%
    mutate(dec_x = form_deciles(x))
df

Description

Produce a table mapping announcements to trading dates.

Usage

get_annc_dates(conn)

Arguments

Value

A tibble with annc_date (each calendar date) and td (the index of that date's trading day, or the next trading day for a non-trading date).

Examples

## Not run: 
# Requires an authenticated PostgreSQL connection with access to CRSP data.
local({
  pg <- DBI::dbConnect(RPostgres::Postgres())
  on.exit(DBI::dbDisconnect(pg), add = TRUE)
  head(get_annc_dates(pg))
})

## End(Not run)

Description

Produce a table of event returns from CRSP.

Usage

get_event_cum_rets(
  data,
  conn,
  permno = "permno",
  event_date = "event_date",
  win_start = 0,
  win_end = 0,
  end_event_date = NULL,
  suffix = ""
)

Arguments

Value

A tibble with one row per security and event window, containing ret_raw (compounded stock return), ret_mkt (stock minus market return), and ret_sz (stock minus size-portfolio return). Returns are decimal fractions; suffix is inserted after ret in these column names.

Examples

## Not run: 
# Requires an authenticated PostgreSQL connection with access to CRSP data.
local({
  pg <- DBI::dbConnect(RPostgres::Postgres())
  on.exit(DBI::dbDisconnect(pg), add = TRUE)
  events <- data.frame(permno = c(14593L, 10107L),
                        event_date = as.Date(c("2019-01-31", "2019-01-31")))
  get_event_cum_rets(events, pg, win_start = -1, win_end = 1)
})

## End(Not run)

Description

Produce a table of event returns from CRSP

Usage

get_event_cum_rets_mth(
  data,
  conn,
  permno = "permno",
  event_date = "event_date",
  win_start = 0,
  win_end = 0,
  end_event_date = NULL,
  suffix = ""
)

Arguments

Value

A tibble with one row per security and event window, containing ret_raw (compounded stock return), ret_mkt (stock minus market return), and ret_sz (stock minus size-portfolio return). Returns are decimal fractions; suffix is inserted after ret in these column names.

Examples

## Not run: 
# Requires an authenticated PostgreSQL connection with access to CRSP data.
local({
  pg <- DBI::dbConnect(RPostgres::Postgres())
  on.exit(DBI::dbDisconnect(pg), add = TRUE)
  events <- data.frame(permno = c(14593L, 10107L),
                        event_date = as.Date(c("2019-01-31", "2019-01-31")))
  get_event_cum_rets_mth(events, pg, win_start = -1, win_end = 1)
})

## End(Not run)

Description

Produce a table of event dates for linking with CRSP.

Usage

get_event_dates(
  data,
  conn,
  permno = "permno",
  event_date = "event_date",
  win_start = 0,
  win_end = 0,
  end_event_date = NULL
)

Arguments

Value

A tibble retaining the security and event-date columns from data, with start_date and end_date giving the mapped trading-window boundaries.

Examples

## Not run: 
# Requires an authenticated PostgreSQL connection with access to CRSP data.
local({
  pg <- DBI::dbConnect(RPostgres::Postgres())
  on.exit(DBI::dbDisconnect(pg), add = TRUE)
  events <- data.frame(permno = c(14593L, 10107L),
                        event_date = as.Date(c("2019-01-31", "2019-01-31")))
  get_event_dates(events, pg, win_start = -1, win_end = 1)
})

## End(Not run)

Description

Produce a table of event returns from CRSP.

Usage

get_event_rets(
  data,
  conn,
  permno = "permno",
  event_date = "event_date",
  win_start = 0,
  win_end = 0,
  end_event_date = NULL
)

Arguments

Value

A tibble containing security and event dates, trading dates, stock returns (ret), benchmark returns (vwretd and decret), window boundaries, and relative_td (trading days from the mapped announcement date).

Examples

## Not run: 
# Requires an authenticated PostgreSQL connection with access to CRSP data.
local({
  pg <- DBI::dbConnect(RPostgres::Postgres())
  on.exit(DBI::dbDisconnect(pg), add = TRUE)
  events <- data.frame(permno = c(14593L, 10107L),
                        event_date = as.Date(c("2019-01-31", "2019-01-31")))
  get_event_rets(events, pg, win_start = -1, win_end = 1)
})

## End(Not run)

Description

Fetch Fama-French industry grouping from Ken French's website.

Usage

get_ff_ind(ind, refresh_cache = FALSE)

Arguments

Value

A tibble with industry identifiers (ff_ind), short and long descriptions (ff_ind_short_desc, ff_ind_desc), SIC interval endpoints (sic_min, sic_max), and SIC interval descriptions (sic_desc).

Examples

# Downloads an industry classification from Ken French's website.
local({
  cache_dir <- tempfile("farr-example-cache-")
  old_options <- options(farr.cache_dir = cache_dir)
  on.exit({
    options(old_options)
    unlink(cache_dir, recursive = TRUE)
  }, add = TRUE)
  head(get_ff_ind(5))
})

Description

Function to generate simulated panel data as described in Gow, Ormazabal and Taylor (2010).

Usage

get_got_data(N = 400, T = 20, Xvol, Evol, rho_X, rho_E)

Arguments

Value

A tibble with columns y (outcome), x (predictor), firm (integer firm identifier), and year (integer year index), with N * T rows.

Source

https://www.jstor.org/stable/20744139

Examples

set.seed(2021)
test <- get_got_data(N = 500, T = 10, Xvol = 0.75,
                     Evol = 0.75, rho_X = 0.5, rho_E = 0.5)

Description

Periods defined by precedent-setting legal cases adopting or rejecting the Inevitable Disclosure Doctrine (IDD) by state.

Usage

get_idd_periods(min_date, max_date)

Arguments

Details

Three kinds of period by state:

• Pre-adoption

• Post-adoption

• Post-rejection

Value

tibble with four columns: state, period_type, start_date, end_date

Examples

idd_periods <- get_idd_periods(min_date = "1994-01-01",
                               max_date = "2010-12-31")
idd_periods

Description

Create a table with cut-offs for size portfolios

Usage

get_me_breakpoints(keep_max = FALSE)

Arguments

Value

A tibble with columns month (a Date), decile (integers 1–10), me_min and me_max (lower and upper market-equity cut-offs in millions of dollars). Unless keep_max is TRUE, the highest upper bound is Inf.

Examples

# Downloads market-equity breakpoints from Ken French's website.
head(get_me_breakpoints())

Description

Create a table of monthly returns for size portfolios

Usage

get_size_rets_monthly()

Value

A tibble with columns month (a Date), decile (integers 1–10), ew_ret (equal-weighted returns), and vw_ret (value-weighted returns). Returns are expressed as decimal fractions.

Examples

# Downloads portfolio returns from Ken French's website.
head(get_size_rets_monthly())

Description

A function returning simulated data on test_scores.

Usage

get_test_scores(
  effect_size = 15,
  n_students = 1000L,
  n_grades = 4L,
  include_unobservables = FALSE,
  random_assignment = FALSE
)

Arguments

Value

A tibble with student identifiers, grades, treatment and post-treatment indicators, and simulated test scores. If include_unobservables is TRUE, it also includes student talent and grade effects.

Examples

set.seed(2021)
library(dplyr, warn.conflicts = FALSE)
get_test_scores() %>% head()

Description

Produce a table mapping dates on CRSP to "trading days". Returned table has two columns: date, a trading date on CRSP; td, a sequence of integers ordered by date.

Usage

get_trading_dates(conn)

Arguments

Value

A tibble with columns date (CRSP trading dates) and td (consecutive integer indices ordered by date).

Examples

## Not run: 
# Requires an authenticated PostgreSQL connection with access to CRSP data.
local({
  pg <- DBI::dbConnect(RPostgres::Postgres())
  on.exit(DBI::dbDisconnect(pg), add = TRUE)
  head(get_trading_dates(pg))
})

## End(Not run)

Description

Link table from GVKEYs to CIKs

Usage

gvkey_ciks

Format

A tibble with 49,089 rows and 5 variables:

gvkey

GVKEY (Compustat firm identifier)

iid

Issue ID

cik

CIK (SEC firm identifier)

first_date

First link date

last_date

Last link date

Description

Dates of precedent-setting legal cases adopting or reject the Inevitable Disclosure Doctrine (IDD) by state.

Usage

idd_dates

Format

A tibble with 24 rows and 3 variables:

state

Two-letter state abbreviation

idd_date

Date of precedent-setting legal case

idd_type

Either "Adopt" or "Reject"

Source

doi:10.1016/j.jfineco.2018.02.008

Description

Data on public float of listed companies from Iliev (2010).

Usage

iliev_2010

Format

A tibble with 7,214 rows and 9 variables:

gvkey

Compustat firm identifier (GVKEY)

fyear

Fiscal year

fdate

Date of end of fiscal year

pfdate

Date for public float value

pfyear

Year for public float value

publicfloat

Public float in $ million

mr

Indicator for filing of a management report

af

Indicator for accelerator filer

cik

SEC firm identifier (CIK)

Details

Five public-float dates have been corrected against the companies' Form 10-K cover pages: Spectrum Control (CIK 92769), fiscal year 2004, and Whitehall Jewellers (CIK 868984), fiscal years 2003–2006. These corrections replace four fractional values of pfyear and correct four months, with corresponding updates to pfdate. The original source CSV is preserved. The original values, corrected dates, and filing links are recorded in data-raw/iliev_2010_date_corrections.csv in the source repository; see also the investigation and filing evidence. Other source anomalies, including invalid dates represented by missing values, remain in the data.

Source

doi:10.1111/j.1540-6261.2010.01564.x

Description

GVKEYs used in Li, Lin and Zhang (2018).

Usage

llz_2018

Format

A tibble with 5,830 rows and 1 variable:

gvkey

GVKEY

Source

doi:10.1111/1475-679X.12187

Description

Function to read data from a parquet file data_dir/schema/table_name.parquet into a table in the DuckDB database at conn.

Usage

load_parquet(conn, table, schema = "", data_dir = Sys.getenv("DATA_DIR"))

Arguments

Value

A lazy dbplyr table backed by conn, with the columns and rows stored in the requested Parquet file. Use dplyr::collect() to retrieve it.

Examples

if (requireNamespace("duckdb", quietly = TRUE)) {
  local({
    conn <- DBI::dbConnect(duckdb::duckdb(),
                           config = list(threads = 2, home_directory = tempdir()))
    on.exit(DBI::dbDisconnect(conn, shutdown = TRUE), add = TRUE)
    data_dir <- tempfile("farr-example-")
    on.exit(unlink(data_dir, recursive = TRUE), add = TRUE)
    data <- dplyr::copy_to(conn, data.frame(id = 1:3, value = c(10, 20, 30)),
                           "example_data")
    saved <- duckdb_to_parquet(data, "example_data", schema = "",
                               data_dir = data_dir)
    dplyr::collect(saved)
    loaded <- load_parquet(conn, "example_data", data_dir = data_dir)
    dplyr::collect(loaded)
  })
}

Description

Data on firms suffering natural disasters based on the sample in Michels (2017).

Usage

michels_2017

Format

A tibble with 423 rows and 12 variables:

cusip

CUSIP supplied by Michels (2017)

eventdate

Date of relevant natural disaster supplied by Michels (2017)

cik

Matched CIK (SEC firm identifier)

permno

Matched PERMNO (CRSP security identifier)

gvkey

Matched GVKEY (Compustat firm identifier)

date_filed

Date of next filing of type 10-Q, 10-K, 10QSB, 10-K405 after event

form_types

List of relevant form types filed on date_filed

next_period_end

Next fiscal period-end after event date

next_fqtr

Fiscal quarter of next period-end after event date

prev_period_end

Last fiscal period-end before event date

prev_fqtr

Fiscal quarter of last period-end before event date

recognize

Indicator for event being recognized (next_period_end before date_filed)

Source

doi:10.1111/1475-679X.12128

Description

A function returning NDCG-at-k metric.

Usage

ndcg(scores, response, k = 0.01)

Arguments

Value

A named numeric vector containing normalized discounted cumulative gain among the highest-scoring fraction k of observations.

Examples

scores <- c(0.1, 0.4, 0.35, 0.8)
response <- c(0, 0, 1, 1)
ndcg(scores, response, k = 0.5)

Description

Function to get data from a table on the WRDS PostgreSQL server and save to local parquet file using DuckDB.

Usage

pg_to_parquet(table_name, schema, data_dir = Sys.getenv("DATA_DIR"))

Arguments

Value

Number of rows created

Examples

## Not run: 
# Requires PostgreSQL credentials for a WRDS database and the duckdb package.
local({
  data_dir <- tempfile("farr-example-")
  dir.create(data_dir)
  on.exit(unlink(data_dir, recursive = TRUE), add = TRUE)
  pg_to_parquet("dsi", "crsp", data_dir = data_dir)
})

## End(Not run)

Description

A function returning data for a ROC plot.

Usage

roc(scores, response)

Arguments

Value

A tibble with columns fpr (false positive rate) and tpr (true positive rate), with one row per distinct point on the ROC curve.

Examples

scores <- c(0.1, 0.4, 0.35, 0.8)
response <- c(0, 0, 1, 1)
roc(scores, response)

Description

Function to create temporary training dataset using distribution implied by w.

Usage

rus(y_train, ir = 1)

Arguments

Details

Following MATLAB, function samples observations of the minority class with replacement and observations of the majority class without replacement. Both classes must be present, with no missing responses. The requested majority sample size must be a positive integer no larger than the majority class; fractional imbalance ratios are allowed if they meet this requirement.

Value

vector

Description

RUSBoost for two-class problems.

Usage

rusboost(formula, df, size, ir = 1, learn_rate = 1, rus = TRUE, control)

## S3 method for class 'rusboost'
predict(object, newdata, type = "prob", ...)

Arguments

Details

A learner that correctly classifies every observation in the full training data stops training successfully. It replaces any earlier learners and is returned as the sole predictive learner with voting weight 1, a finite representation of its limiting vote. Its observation-weight vector is the normalized vector used for that attempt.

Learners with weighted error at least one half receive no vote. AdaBoost (rus = FALSE) stops at such a learner; RUSBoost skips it and tries another sample, up to size attempts. Training raises an error if no usable learner is found. The response must contain two nonmissing classes, including class 1.

Value

rusboost() returns an object of class rusboost, a list containing fitted trees (weakLearners), their voting weights (alpha), observation-weight vectors (w), and the training formula. predict() returns a numeric vector of weighted positive-vote shares for type = "prob", or classes -1 and 1 (0 for a tied vote) for type = "class".

Examples

training <- data.frame(x = 1:8, y = factor(c(rep(0, 4), rep(1, 4))))
control <- rpart::rpart.control(minsplit = 2, minbucket = 1, cp = 0, xval = 0)
fit <- rusboost(y ~ x, training, size = 3, rus = FALSE, control = control)
predict(fit, training, type = "prob")
predict(fit, training, type = "class")

Description

A data set containing the tickers and company names for Russell 3000 at time SEC created the pilot sample. Data are created from sample supplied by FHK.

Usage

sho_r3000

Format

A tibble with 3000 rows × 2 variables.

russell_ticker

Ticker

russell_name

Company name

Source

doi:10.1111/jofi.12369

Description

A data set containing the tickers, PERMNOs, GVKEYs, and treatment assignments for Russell 3000 sample used by SEC.

Usage

sho_r3000_gvkeys

Format

A tibble with 2,951 rows × 3 variables.

ticker

Ticker

permno

PERMNO (CRSP security identifier)

gvkey

GVKEY (Compustat firm identifier)

pilot

Indicator for stock being part of Reg SHO pilot program

Source

https://iangow.github.io/far_book/natural-revisited.html#the-sho-pilot-sample

Description

A data set containing the tickers, PERMNOs, and treatment assignments for Russell 3000 sample used by SEC.

Usage

sho_r3000_sample

Format

A tibble with 2,954 rows × 3 variables.

ticker

Ticker

permno

PERMNO (CRSP security identifier)

pilot

Indicator for stock being part of Reg SHO pilot program

Source

https://iangow.github.io/far_book/natural-revisited.html#the-sho-pilot-sample

Description

A data set containing the tickers and company names for pilot firms from Reg SHO pilot. Data are scraped from the SEC's own website.

Usage

sho_tickers

Format

A tibble with 986 rows × 2 variables.

ticker

Ticker

co_name

Company name

Source

https://www.sec.gov/rule-release/34-50104

Description

Data on firm headquarters based on SEC EDGAR filings. Dates related to SEC filing dates. Rather than provide dates for all filings, data are aggregated into groups of filings by state and CIK and dates are collapsed into windows over which all filings for a given CIK were associated with a given state. For example, CIK 0000037755 has filings with a CA headquarters from 1994-06-02 until 1996-03-25, then filings with an OH headquarters from 1996-05-30 until 1999-04-05, then filings with a CA headquarters from 1999-06-11 onwards. To ensure continuous coverage over the sample period, it is assumed that any change in state occurs the day after the last observed filing for the previous state.

Usage

state_hq

Format

A tibble with 53,133 rows and 4 variables:

cik

SEC's Central Index Key (CIK)

ba_state

Two-letter abbreviation of state

min_date

Date of first filing with CIK-state combination in a contiguous series of filings

max_date

Date of last filing with CIK-state combination in a contiguous series of filings

Source

https://sraf.nd.edu/sec-edgar-data/10-x-header-data/

Description

Print CPU (and other) times that expr used, return value of expr.

Usage

system_time(expr)

Arguments

Value

Result of evaluating expr

Examples

result <- system_time(sum(1:10))
result

Description

A simulated data set of test scores.

Usage

test_scores

Format

A tibble with 4,000 rows and 3 variables:

id

Student identifier

grade

School grade at time of test

score

Test score

Description

Truncate a vector at prob and 1 - prob. Extreme values are turned into NA values.

Usage

truncate(x, prob = 0.01, p_low = prob, p_high = 1 - prob)

Arguments

Value

A vector the same length as x, with values outside the specified quantiles replaced by NA.

Examples

trunced <- truncate(1:100, prob = 0.05)
min(trunced, na.rm = TRUE)
max(trunced, na.rm = TRUE)

Description

Data to be combined with data in compsegd.seg_customer to create an indicator for non-disclosure of customer names.

Usage

undisclosed_names

Format

A tibble with 460 rows and 2 variables:

cnms

Matches field in compsegd.seg_customer (WRDS)

disclosure

Indicator that name is not disclosed

Description

Winsorize a vector at prob and 1 - prob.

Usage

winsorize(x, prob = 0.01, p_low = prob, p_high = 1 - prob)

Arguments

Value

A vector the same length as x, with values outside the specified quantiles replaced by the corresponding lower or upper cut-off.

Examples

winsorized <- winsorize(1:100, prob = 0.05)
min(winsorized, na.rm = TRUE)
max(winsorized, na.rm = TRUE)

Description

A data set containing the event dates used in Zhang (2007). Data obtained from Panel of Table of Zhang (2007). If an event spans multiple dates, then a row is included for each date.

Usage

zhang_2007_events

Format

A tibble with 30 rows × 3 variables.

event

Identifier for the event

date

Date of event

event_desc

Description of the event

Source

doi:10.1016/j.jacceco.2007.02.002

Description

A data set containing the event windows used in Zhang (2007). Data obtained from Panel of Table of Zhang (2007).

Usage

zhang_2007_windows

Format

A tibble with 17 rows × 3 variables.

event

Identifier for the event

beg_date

First date of event window

end_date

Last date of event window

Source

doi:10.1016/j.jacceco.2007.02.002