| Title: | Algorithms for Matching Bullet Lands |
|---|---|
| Description: | Analyze bullet lands using nonparametric methods. We provide a reading routine for x3p files (see <http://www.openfmc.org> for more information) and a host of analysis functions designed to assess the probability that two bullets were fired from the same gun barrel. |
| Authors: | Eric Hare [aut, cre], Heike Hofmann [aut], Ganesh Krishnan [aut] |
| Maintainer: | Eric Hare <[email protected]> |
| License: | MIT |
| Version: | 0.1.0.9003 |
| Built: | 2025-02-25 03:00:53 UTC |
| Source: | https://github.com/erichare/bulletr |
Fit a LOESS model with bootstrap samples
boot_fit_loess(bullet, groove, B = 1000, alpha = 0.95)boot_fit_loess(bullet, groove, B = 1000, alpha = 0.95)
bullet |
Bullet as returned from x3p_to_df |
groove |
Groove as returned from get_grooves |
B |
number of Bootstrap samples |
alpha |
The significance level |
Some more info - not sure at the moment which bullet this is. Describe structure.
br411br411
a list
The bullet with the first name serves as a reference, the second bullet is shifted.
bulletAlign(data, value = "l30")bulletAlign(data, value = "l30")
data |
data frame consisting of at least two surface crosscuts as given by function |
value |
string of the variable to match. Defaults to l30, the variable returned from function |
list consisting of a) the maximal cross correlation, b) the lag resulting in the highest cross correlation, and c) same data frame as input, but y vectors are aligned for maximal correlation
Identifies a "representative" cross section for a bullet land engraved area. Striation marks on a bullet land are the best expressed at the heel (bottom) of a bullet where break-off is still problematic. Using cross-correlation we identify a cross section that is the closest to the bottom of the bullet but does not suffer from break-off. If the resulting cross section is equal to the maximum of the search area (defined in xlimits), there should be some investigation, whether this cross section is usable. There is the risk of tank rash. XXX still to do: are missing values only on the right hand side (leading shoulder)?
bulletCheckCrossCut(path, bullet = NULL, distance = 25, xlimits = c(50, 500), minccf = 0.9, span = 0.03, percent_missing = 50)bulletCheckCrossCut(path, bullet = NULL, distance = 25, xlimits = c(50, 500), minccf = 0.9, span = 0.03, percent_missing = 50)
path |
path to an x3p file. Ignored in case bullet is not NULL. |
bullet |
if not NULL, the bullet land engraved area (in x3p format). |
distance |
positive numeric value indicating the distance between cross sections to use for a comparison |
xlimits |
vector of values between which to check for cross sections in a stable region |
minccf |
minimal value of cross correlation to indicate a stable region |
span |
The span for the loess smooth function |
percent_missing |
maximum percent missing values on the crosscut to be picked |
Identify the number of maximum CMS between two bullet lands
bulletGetMaxCMS(lof1, lof2, column = "resid", span = 35)bulletGetMaxCMS(lof1, lof2, column = "resid", span = 35)
lof1 |
dataframe of smoothed first signature |
lof2 |
dataframe of smoothed second signature |
column |
The column which to smooth |
span |
positive number for the smoothfactor to use for assessing peaks. |
list of matching parameters, data set of the identified striae, and the aligned data sets.
Identify the number of maximum CMS between two bullet lands
bulletGetMaxCMS_nist(lof1, lof2, column = "resid", span = 35)bulletGetMaxCMS_nist(lof1, lof2, column = "resid", span = 35)
lof1 |
dataframe of smoothed first signature |
lof2 |
dataframe of smoothed second signature |
column |
The column which to smooth |
span |
positive number for the smoothfactor to use for assessing peaks. |
list of matching parameters, data set of the identified striae, and the aligned data sets.
Smooth the surface of a bullet
bulletSmooth(data, span = 0.03, limits = c(-5, 5), id = "bullet")bulletSmooth(data, span = 0.03, limits = c(-5, 5), id = "bullet")
data |
data frame as returned by the function |
span |
width of the smoother, defaults to 0.03 |
limits |
vector of the form c(min, max). Results will be limited to be between these values. |
id |
variable name of the land identifier |
data frame of the same form as the input extended by the vector l30 for the smooth.
Chumbley test score
chumbley(b1, b2, window, reps = 3)chumbley(b1, b2, window, reps = 3)
b1 |
dataframe |
b2 |
dataframe |
window |
width of the window (in indices) to consider for matching |
reps |
number of replicates to use in the evaluation |
library(dplyr) data(br411) b1 <- get_crosscut(x = 250, bullet=br411) b2 <- get_crosscut(x = 150, bullet = br411) b3 <- get_crosscut(x = 10, bullet=br411) b1.gr <- b1 %>% get_grooves(smoothfactor=30) b2.gr <- b2 %>% get_grooves() b3.gr <- b3 %>% get_grooves() # get signatures b1 <- fit_loess(b1, b1.gr)$data b2 <- fit_loess(b2, b2.gr)$data b3 <- fit_loess(b3, b3.gr)$data match12 <- get_lag_max_R(b1, b2, window = 100, b1.left = 450) # matched correlations get_cor(b1, b2, window = 100, b1.left = 800, lag = match12$lag) get_cor(b1, b2, window = 100, b1.left = 1000, lag = match12$lag) get_cor(b1, b2, window = 100, b1.left = 1200, lag = match12$lag) # random correlations get_cor(b1, b2, window = 100, b1.left = 800, lag = 100) get_cor(b1, b2, window = 100, b1.left = 1000, lag = -300) get_cor(b1, b2, window = 100, b1.left = 1200, lag = -500) chumbley(b1, b2, window=150, reps=5) match13 <- get_lag_max_R(b1, b3, window = 100, b1.left = 450) # matched correlations get_cor(b1, b3, window = 100, b1.left = 800, lag = match13$lag) get_cor(b1, b3, window = 100, b1.left = 1000, lag = match13$lag) get_cor(b1, b3, window = 100, b1.left = 1200, lag = match13$lag) # random correlations get_cor(b1, b3, window = 100, b1.left = 800, lag = 100) get_cor(b1, b3, window = 100, b1.left = 1000, lag = 300) get_cor(b1, b3, window = 100, b1.left = 1200, lag = 500) chumbley(b1, b3, window=100, reps=5)library(dplyr) data(br411) b1 <- get_crosscut(x = 250, bullet=br411) b2 <- get_crosscut(x = 150, bullet = br411) b3 <- get_crosscut(x = 10, bullet=br411) b1.gr <- b1 %>% get_grooves(smoothfactor=30) b2.gr <- b2 %>% get_grooves() b3.gr <- b3 %>% get_grooves() # get signatures b1 <- fit_loess(b1, b1.gr)$data b2 <- fit_loess(b2, b2.gr)$data b3 <- fit_loess(b3, b3.gr)$data match12 <- get_lag_max_R(b1, b2, window = 100, b1.left = 450) # matched correlations get_cor(b1, b2, window = 100, b1.left = 800, lag = match12$lag) get_cor(b1, b2, window = 100, b1.left = 1000, lag = match12$lag) get_cor(b1, b2, window = 100, b1.left = 1200, lag = match12$lag) # random correlations get_cor(b1, b2, window = 100, b1.left = 800, lag = 100) get_cor(b1, b2, window = 100, b1.left = 1000, lag = -300) get_cor(b1, b2, window = 100, b1.left = 1200, lag = -500) chumbley(b1, b2, window=150, reps=5) match13 <- get_lag_max_R(b1, b3, window = 100, b1.left = 450) # matched correlations get_cor(b1, b3, window = 100, b1.left = 800, lag = match13$lag) get_cor(b1, b3, window = 100, b1.left = 1000, lag = match13$lag) get_cor(b1, b3, window = 100, b1.left = 1200, lag = match13$lag) # random correlations get_cor(b1, b3, window = 100, b1.left = 800, lag = 100) get_cor(b1, b3, window = 100, b1.left = 1000, lag = 300) get_cor(b1, b3, window = 100, b1.left = 1200, lag = 500) chumbley(b1, b3, window=100, reps=5)
Table of the number of consecutive matches
CMS(match)CMS(match)
match |
is a Boolean vector of matches/non-matches |
a table of the number of the CMS and their frequencies
x <- rbinom(100, size = 1, prob = 1/3) CMS(x == 1) # expected value for longest match is 3x <- rbinom(100, size = 1, prob = 1/3) CMS(x == 1) # expected value for longest match is 3
Get average scores
compute_average_scores(land1, land2, score)compute_average_scores(land1, land2, score)
land1 |
(numeric) vector with land ids of bullet 1 |
land2 |
(numeric) vector with land ids of bullet 2 |
score |
numeric vector of scores to be summarized |
numeric vector of average scores. Length is the same as the number of land engraved areas on the bullets.
The first vector serves as a reference, the second vector is shifted, such that it aligns best with the first and has the same length as the first vector.
do_align(y1, y2, min.overlap = 0.1 * max(length(y1), length(y2)))do_align(y1, y2, min.overlap = 0.1 * max(length(y1), length(y2)))
y1 |
vector of striation marks (assuming equidistance between values) |
y2 |
vector of striation marks |
min.overlap |
integer value: what is the minimal number of values between y1 and y2 that should be considered? |
list consisting of a) the maximal cross correlation, b) the lag resulting in the highest cross correlation, and c) a vector of length y1 with aligned values of y2.
library(dplyr) x <- runif(20) do_align(x, lead(x, 5)) do_align(x, lag(x, 5), min.overlap=2) do_align(x, lag(x, 5), min.overlap=3) do_align(x, x[-(1:5)], min.overlap=3) do_align(x[-(1:5)], x, min.overlap=3)library(dplyr) x <- runif(20) do_align(x, lead(x, 5)) do_align(x, lag(x, 5), min.overlap=2) do_align(x, lag(x, 5), min.overlap=3) do_align(x, x[-(1:5)], min.overlap=3) do_align(x[-(1:5)], x, min.overlap=3)
First, the surface measurements of the bullet land is trimmed to be within left and right groove as specified by vector groove.
A loess regression is fit to the remaining surface measurements and residuals are calculated.
The most extreme 0.25
The result is called the signature of the bullet land.
fit_loess(bullet, groove, span = 0.75)fit_loess(bullet, groove, span = 0.75)
bullet |
The bullet object as returned from x3p_to_df |
groove |
vector of two numeric values indicating the location of the left and right groove. |
span |
The span to use for the loess regression |
a list of a data frame of the original bullet measurements extended by loess fit, residuals, and standard errors and two plots: a plot of the fit, and a plot of the bullet's land signature.
old function - for previous ISO standard. x3p format consists of a list with header info and a 2d matrix of scan depths. fortify_x3p turn the matrix into a variable within a data frame, using the parameters of the header as necessary.
fortify_x3p(x3p)fortify_x3p(x3p)
x3p |
a file in x3p format as return by function read_x3p |
data frame with variables x, y, and value
data(br411) br411_fort <- fortify_x3p(br411) head(br411_fort)data(br411) br411_fort <- fortify_x3p(br411) head(br411_fort)
Deprecated function use get_crosscut
get_bullet(path, x = 243.75)get_bullet(path, x = 243.75)
path |
The path to the x3p file |
x |
The crosscut value |
Compute a Chumbley test score
get_chumbley(y1, y2, window, reps = 3)get_chumbley(y1, y2, window, reps = 3)
window |
width of the window (in indices) to consider for matching |
reps |
number of replicates to use in the evaluation |
b1 |
vector of equi-distant toolmark values |
b2 |
vector of equi-distant toolmark values |
Chumbley, L. S., Morris, M. D., Kreiser, M. J., Fisher, C., Craft, J., Genalo, L. J., Davis, S., Faden, D. and Kidd, J. (2010), Validation of Tool Mark Comparisons Obtained Using a Quantitative, Comparative, Statistical Algorithm. Journal of Forensic Sciences, 55: 953–961. doi:10.1111/j.1556-4029.2010.01424.x
library(dplyr) data(br411) b1 <- get_crosscut(x = 250, bullet=br411) b2 <- get_crosscut(x = 150, bullet = br411) b3 <- get_crosscut(x = 10, bullet=br411) b1.gr <- b1 %>% get_grooves(smoothfactor=30) b2.gr <- b2 %>% get_grooves() b3.gr <- b3 %>% get_grooves() # check that the grooves are actually found: b1.gr$plot b2.gr$plot # get signatures b1 <- fit_loess(b1, b1.gr)$data b2 <- fit_loess(b2, b2.gr)$data b3 <- fit_loess(b3, b3.gr)$data get_chumbley(b1$resid, b2$resid, window=150, reps=5) get_chumbley(b1$resid, b2$resid, window=150, reps=5) get_chumbley(b1$resid, b2$resid, window=50, reps=12)library(dplyr) data(br411) b1 <- get_crosscut(x = 250, bullet=br411) b2 <- get_crosscut(x = 150, bullet = br411) b3 <- get_crosscut(x = 10, bullet=br411) b1.gr <- b1 %>% get_grooves(smoothfactor=30) b2.gr <- b2 %>% get_grooves() b3.gr <- b3 %>% get_grooves() # check that the grooves are actually found: b1.gr$plot b2.gr$plot # get signatures b1 <- fit_loess(b1, b1.gr)$data b2 <- fit_loess(b2, b2.gr)$data b3 <- fit_loess(b3, b3.gr)$data get_chumbley(b1$resid, b2$resid, window=150, reps=5) get_chumbley(b1$resid, b2$resid, window=150, reps=5) get_chumbley(b1$resid, b2$resid, window=50, reps=12)
Get correlation between two signatures
get_cor(b1, b2, window, b1.left, lag)get_cor(b1, b2, window, b1.left, lag)
b1 |
dataframe |
b2 |
dataframe |
window |
width of the window (in indices) to consider for matching |
b1.left |
left index location of the matching window |
lag |
integer lag for the second window |
Read a crosscut from a 3d surface file
get_crosscut(path = NULL, x = 243.75, bullet = NULL)get_crosscut(path = NULL, x = 243.75, bullet = NULL)
path |
path to an x3p file. The path will only be considered, if bullet is not specified. |
x |
level of the crosscut to be taken. If this level does not exist, the crosscut with the closest level is returned. |
bullet |
alternative access to the surface measurements. |
data frame
Compute the Euclidean distance between two toolmark patterns. The striation patterns are not aligned before the distance is calculated.
get_D(y1, y2, normalize = TRUE, resolution = 0.645)get_D(y1, y2, normalize = TRUE, resolution = 0.645)
y1 |
vector of equi-distant toolmark values |
y2 |
vector of equi-distant toolmark values |
normalize |
should the result be normalized to 1000 microns (1 millimeter)? Defaults to TRUE. |
resolution |
microns per pixel. Only used for normalization. |
Get a feature vector for a pair of lands
get_features(res)get_features(res)
res |
list of two aligned lands resulting from bulletGetMaxCMS |
Find the grooves of a bullet land
get_grooves(bullet, method = "rollapply", smoothfactor = 15, adjust = 10, groove_cutoff = 400, mean_left = NULL, mean_right = NULL, mean_window = 100)get_grooves(bullet, method = "rollapply", smoothfactor = 15, adjust = 10, groove_cutoff = 400, mean_left = NULL, mean_right = NULL, mean_window = 100)
bullet |
data frame with topological data in x-y-z format |
method |
method to use for identifying grooves. Defaults to "rollapply" |
smoothfactor |
The smoothing window to use - XXX the smoothing window seems to depend on the resolution at which the data has been collected. |
adjust |
positive number to adjust the grooves - XXX should be expressed in microns rather than an index |
groove_cutoff |
The index at which a groove cannot exist past - XXX this parameter should be expressed in microns rather than as an index to be able to properly deal with different resolutions |
mean_left |
If provided, the location of the average left groove |
mean_right |
If provided, the location of the average right groove |
mean_window |
The window around the means to use |
second_smooth |
Whether or not to smooth a second time |
Use the center of a crosscut
get_grooves_middle(bullet, middle = 75)get_grooves_middle(bullet, middle = 75)
bullet |
data frame with topological data in x-y-z format |
middle |
middle percent to use for the identification |
Use a robust fit of a quadratic curve to find groove locations
get_grooves_quadratic(bullet, adjust)get_grooves_quadratic(bullet, adjust)
bullet |
data frame with topological data in x-y-z format |
adjust |
positive number to adjust the grooves |
Using rollapply to find grooves in a crosscut
get_grooves_rollapply(bullet, smoothfactor = 15, adjust = 10, groove_cutoff = 400, mean_left = NULL, mean_right = NULL, mean_window = 100, second_smooth = T, which_fun = mean)get_grooves_rollapply(bullet, smoothfactor = 15, adjust = 10, groove_cutoff = 400, mean_left = NULL, mean_right = NULL, mean_window = 100, second_smooth = T, which_fun = mean)
bullet |
data frame with topological data in x-y-z format |
smoothfactor |
The smoothing window to use |
adjust |
positive number to adjust the grooves |
groove_cutoff |
The index at which a groove cannot exist past |
mean_left |
If provided, the location of the average left groove |
mean_right |
If provided, the location of the average right groove |
mean_window |
The window around the means to use |
second_smooth |
Whether or not to smooth a second time |
which_fun |
Which function to use in the rollapply statement |
Compute the Housdorff distance between two toolmark patterns. The striation patterns are not aligned before the distance is calculated. The Hausdorff distance is defined as the maximum among the shortest distances between two curves. Here, we allow to trim the largest distances to make the distance more robust
get_H(y1, y2, trim = 0)get_H(y1, y2, trim = 0)
y1 |
vector of equi-distant toolmark values |
y2 |
vector of equi-distant toolmark values |
trim |
percentage of largest distances to be trimmed. |
A small piece (b2) is matched to a much larger piece (b1). The lag gives the index location of the best match of b2 in b1.
This function is essentially just a wrapper for my_ccf, but adds a plot of the result for convenience.
get_lag(b1, b2, negperc = 10)get_lag(b1, b2, negperc = 10)
b1 |
vector of striation marks (assuming equidistance between values) |
b2 |
(smaller) vector of striation marks |
negperc |
amount of lead that b2 can have compared to b1 |
list of lag and correlation achieved. The plot shows b2 on b1
See Chumbley et al (2010). A small piece (b2) is matched to a much larger piece (b1). The lag gives the index location of the best match of b2 in b1.
get_lag_max_R(b1, b2, window, b1.left)get_lag_max_R(b1, b2, window, b1.left)
b1 |
dataframe |
b2 |
dataframe |
window |
width of the window (in indices) to consider for matching |
b1.left |
left index of the matching window |
list of lag and correlation achieved. The plot shows b2 on b1.
Identify the location and the depth of peaks and heights at a crosscut
get_peaks(loessdata, column = "resid", smoothfactor = 35, striae = TRUE, window = TRUE)get_peaks(loessdata, column = "resid", smoothfactor = 35, striae = TRUE, window = TRUE)
loessdata |
export from rollapply |
column |
The column which should be smoothed |
smoothfactor |
set to default of 35. Smaller values will pick up on smaller changes in the crosscut. |
striae |
If TRUE, show the detected striae on the plot |
window |
If TRUE, show the window of the striae on the plot |
list of several objects:
Identify the location and the depth of peaks and heights at a crosscut
get_peaks_nist(loessdata, column = "resid", smoothfactor = 35, striae = TRUE, window = TRUE)get_peaks_nist(loessdata, column = "resid", smoothfactor = 35, striae = TRUE, window = TRUE)
loessdata |
export from rollapply |
column |
The column which should be smoothed |
smoothfactor |
set to default of 35. Smaller values will pick up on smaller changes in the crosscut. |
striae |
If TRUE, show the detected striae on the plot |
window |
If TRUE, show the window of the striae on the plot |
list of several objects:
Assuming the variables x and y are describing points located on a circle, the function uses a likelihood approach to estimate center and radius of the circle.
getCircle(x, y)getCircle(x, y)
x |
numeric vector of values |
y |
numeric vector of values |
three dimensional vector of the circle center (x0, y0) and the radius
Estimation of the twist in a barrel follows roughly the process described by Chu et al (2010). At the moment, twist is estimated from a single land - but the twist should be the same for the whole barrel. Therefore all lands of the same barrel should have the same twist. A note on timing: at the moment calculating the twist rate for a bullet land takes several minutes. XXX TODO XXX make the different methods a parameter. Also, accept other input than the path - if we start with the flattened bulletland we get results much faster.
getTwist(path, bullet = NULL, twistlimit = NULL, cutoff = 0.75)getTwist(path, bullet = NULL, twistlimit = NULL, cutoff = 0.75)
path |
to a file in x3p format |
bullet |
data in x3p format as returned by function read_x3p |
twistlimit |
Constraint the possible twist value |
cutoff |
Use this for the quantile cutoff |
numeric value estimating the twist
## Not run: # execution takes several minutes load("data/b1.rda") twist <- getTwist(path="barrel 1 bullet 1", bullet = b1, twistlimit=c(-2,0)*1.5625) ## End(Not run)## Not run: # execution takes several minutes load("data/b1.rda") twist <- getTwist(path="barrel 1 bullet 1", bullet = b1, twistlimit=c(-2,0)*1.5625) ## End(Not run)
Number of maximum consecutively matching striae
maxCMS(match)maxCMS(match)
match |
is a Boolean vector of matches/non-matches |
an integer value of the maximum number of consecutive matches
x <- rbinom(100, size = 1, prob = 1/3) CMS(x == 1) # expected value for longest match is 3 maxCMS(x==1)x <- rbinom(100, size = 1, prob = 1/3) CMS(x == 1) # expected value for longest match is 3 maxCMS(x==1)
Cross correlation function between two vectors
my_ccf(x, y, min.overlap = 0.1 * max(length(x), length(y)))my_ccf(x, y, min.overlap = 0.1 * max(length(x), length(y)))
x |
vector |
y |
vector |
min.overlap |
integer value: what is the minimal number of values between x and y that should be considered? |
list with ccf values and lags
library(dplyr) x <- runif(20) my_ccf(x, lead(x, 5)) my_ccf(x, lag(x, 5), min.overlap=3) x <- runif(100) my_ccf(x[45:50], x, min.overlap=6)library(dplyr) x <- runif(20) my_ccf(x, lead(x, 5)) my_ccf(x, lag(x, 5), min.overlap=3) x <- runif(100) my_ccf(x[45:50], x, min.overlap=6)
Plot a bullet land using plotly
plot_3d_land(path, bullet = NULL, sample = 1, ...)plot_3d_land(path, bullet = NULL, sample = 1, ...)
path |
The path to the x3p file |
bullet |
If not null, use this pre-loaded bullet |
sample |
integer value. take every 1 in sample values from the surface matrix |
... |
parameters passed on to plot_ly call |
data(br411) plot_3d_land(bullet=br411, sample=2)data(br411) plot_3d_land(bullet=br411, sample=2)
estimate a circle, find predictive values and resiudals. depending on specification, vertical (regular) residuals or orthogonal residuals are computed.
predCircle(x, y, resid.method = "response")predCircle(x, y, resid.method = "response")
x |
vector of numeric values |
y |
vector of numeric values |
resid.method |
character, one of "response" or "ortho"(gonal) |
data frame with predictions and residuals
Fit a smooth line throught x and y, find predictive values and resiudals.
predSmooth(x, y)predSmooth(x, y)
x |
vector of numeric values |
y |
vector of numeric values |
data frame with predictions and residuals
x3p file of a 3d topological bullet surface is processed at surface crosscut x,
the bullet grooves in the crosscuts are identified and removed, and a loess smooth
is used (see ?loess for details) to remove the big structure.
x3p file of a 3d topological bullet surface is processed at surface crosscut x,
the bullet grooves in the crosscuts are identified and removed, and a loess smooth
is used (see ?loess for details) to remove the big structure.
processBullets(bullet, name = "", x = 100, grooves = NULL, span = 0.75, window = 1, ...) processBullets(bullet, name = "", x = 100, grooves = NULL, span = 0.75, window = 1, ...)processBullets(bullet, name = "", x = 100, grooves = NULL, span = 0.75, window = 1, ...) processBullets(bullet, name = "", x = 100, grooves = NULL, span = 0.75, window = 1, ...)
bullet |
file as returned from read_x3p |
name |
name of the bullet |
x |
(vector) of surface crosscuts to process (in meters). |
grooves |
The grooves to use as a two element vector, if desired |
span |
The span for the loess fit |
window |
The mean window around the ideal crosscut |
... |
Additional arguments, passed to the get_grooves function |
bullet |
file as returned from read_x3p |
name |
name of the bullet |
x |
(vector) of surface crosscuts to process. |
grooves |
The grooves to use as a two element vector, if desired |
span |
The span for the loess fit |
window |
The window around the ideal crosscut |
... |
Additional arguments, passed to the get_grooves function |
data frame
data frame
data(br411) br411_processed <- processBullets(br411, name = "br411") data(br411) br411_processed <- processBullets(br411, name = "br411")data(br411) br411_processed <- processBullets(br411, name = "br411") data(br411) br411_processed <- processBullets(br411, name = "br411")
Read a dat file and create an x3p file
read_dat(path, profiley = TRUE, sample = 1)read_dat(path, profiley = TRUE, sample = 1)
path |
The file path to the dat file |
profiley |
are profiles on y? |
sample |
1 in sample lines will be taken |
list with header information and surface matrix
## Not run: d1 <- read_dat("Br4 Bullet 4-1.dat") d2 <- read_dat("L1.dat", profiley = FALSE) ## End(Not run)## Not run: d1 <- read_dat("Br4 Bullet 4-1.dat") d2 <- read_dat("L1.dat", profiley = FALSE) ## End(Not run)
this randomforest was fitted to predict known matches and non-matches from the scans of land engraved areas of the Hamby study.
rtreesrtrees
a random forest object fitted by the randomforest function from the package of the same name
Predict smooth from a fit
smoothloess(x, y, span, sub = 2)smoothloess(x, y, span, sub = 2)
x |
X values to use |
y |
Y values to use |
span |
The span of the loess fit |
sub |
Subsample factor |
Match striation marks across two cross sections based on previously identified peaks and valleys
striation_identify(lines1, lines2)striation_identify(lines1, lines2)
lines1 |
data frame as returned from get_peaks function. data frames are expected to have the following variables: xmin, xmax, group, type, bullet, heights |
lines2 |
data frame as returned from get_peaks function. data frames are expected to have the following variables: xmin, xmax, group, type, bullet, heights |
data frame of the same form as lines1 and lines2, but consisting of an additional variable of whether the striation marks are matches
Convert a data frame into an x3p file
unfortify_x3p(df)unfortify_x3p(df)
df |
A data frame produced by fortify_x3p |
An x3p object
data(br411) br411_fort <- fortify_x3p(br411) br411_unfort <- unfortify_x3p(br411_fort) identical(br411_unfort, br411)data(br411) br411_fort <- fortify_x3p(br411) br411_unfort <- unfortify_x3p(br411_fort) identical(br411_unfort, br411)