freq_map: Function for frequency contribution mapping in the real...

In Chitran1987/StatsChitran: StatsChitran

Function for frequency contribution mapping in the real functional domain

Description

The function
1. divides the data(X,Y) into segments decided by either nbin or xbox. For ex. no_of_segments = (max(X) - min(X))/xbox
2. Performs a fourier transform on each of these segments and then integrates between the frequency limits/windows (mentioned in each w_int column)
3. Saves these integration values against each segment number and builds dataframes like data(X, frequency_window(w))
4. Obviously, the no. of such dataframes depends on the no. of such frequency windows asked to integrate between. (This is equal to the no. of columns of the w_int dataframe)
5. Return each such dataframe as a separate object in a list
6. Oviously the length of the returned list is equal to the no. of columns in w_int
7. For plotting, using 'plt=T' and 'plt.leg=T/F' overlays each frequency map available in the returned list with the plotted data(X,Y). 'plt.leg' simply toggles the legend on or off.
8. If 'plt=T', the 'color' vector must be provided with the right required sequence, else function throws an error.

Usage

freq_map(X, Y, w_int, xbox)
freq_map(X,Y, w_int, nbin)
freq_map(X,Y, w_int, xbox, color, plt=T, plt.leg=T)
freq_map(X,Y, w_int, xbox, color, plt=T, plt.leg=F)
freq_map(X,Y, w_int, nbin, color, plt=T, plt.leg=T)
freq_map(X,Y, w_int, nbin, color, plt=T, plt.leg=F)

Arguments

X	A vector with the x-values of the dataset. This argument must be provided.
Y	A vector with the y-values of the dataset. This argument must be provided.
w_int	1. A dataframe with n-columns and 2 rows 2. Each value in w_int is of of 'w = 2*pi*f' or circular frequency format. DONOT use the 'w = 1*f' format 3. Here the no. of required frequency windows is n 4. Each column denotes a given frequency window 5. Each column MUST have EXACTLY 2 values (denoting the upper/lower frequency limit of that window), else error is thrown 6. This argument must be provided
xbox	1. The width of the segmentation box in real space 2. Must be a single number (numeric and 'dim(xbox)=NULL') 3. Must only be provided when 'nbin' is NOT provided 4. If provided along with 'nbin', function throws an error
nbin	1. The no. of periodic units wanted in each freq. window (mapped w.r.t lower limit of freq window) 2. Must be a numeric vector(NOT list) of length = n, otherwise error thrown 3. Must only be provided when 'xbox' is NOT provided 4. If provided along with 'xbox', function throws an error
plt	1. Toggles plot on or off with TRUE/FALSE respectively 2. Default set to FALSE
plt.leg	1. Toggles plot legend on or off with TRUE/FALSE respectively 2. Ordinarily set to FALSE 3. Can only be provided if 'plt=TRUE'
color	1. Vector of length n 2. Provides colours of the different frequency windows to be plotted 3. Must only be provided, if 'plt=TRUE'

Details

1. X, Y, w_int and (either xbox or nbin) are mandatory
2. If 'plt = TRUE', 'color' is mandatory
3. w_int uses 'w = 2*pi*f' format, not 'w = 1*f' format

Value

Returns a list containing dataframes related to each individual frequency window presented in the w_int dataframe

Note

1. If integration errors with "too few samples for integration" are encountered, either increase 'nbin' till the error goes away, or increase window size (space between the w_max and w_min) for each window (column) in 'w_int'. Typically for w_max >= 5*w_min, the integration should work properly

Author(s)

Chitran Ghosal

Examples

### define the w_int dataframe
w1 <- c(550, 850)
w2 <- c(3000, 3300)
w3 <- c(4700, 5000)
w_df <- data.frame(w1, w2, w3) #input for the 'w_int' dataframe

### define the colour vector
col_vec <- c('red', rgb(0,0,1,0.75), rgb(0,1,0,0.5)) #input for the 'color' argument

### define the bin size
bin <- 0.085 #input for 'xbox' argument

### plot the frequency windows
dft <- ft(my_voice_01$time, my_voice_01$left, w = T, set.nyquist = T) #fourier transform my voice

subplot(c(1, 2)) #divide graph area for plotting
plot(dft$wf, abs(dft$fy), type = 'l', col = 'grey', xlab = 'w', ylab = 'amp', main = 'frequency windows/fourier transform', xlim = c(-6000, 6000)) # plot fourier transform
grid(nx = NULL, col = 'black') # grid the fourier transform
legend('topright', legend = names(w_df), lty = 1, col = col_vec, horiz = F, bg = 'transparent', bty = 'n', lwd = 3) # legend
abline(v=c(-w1, w1), col = col_vec[1], lwd = 1.5) # plot the w1 windown
abline(v=c(-w2, w2), col = col_vec[2], lwd = 1.5) # plot the w2 window
abline(v=c(-w3, w3), col = col_vec[3], lwd = 1.5) # plot the w3 window

### calculate/plot the freq_map of the windows w1, w2 and w3
List_frmap <- freq_map(my_voice_01$time, my_voice_01$left, w_int = w_df, xbox = bin, plt = T, plt.leg = T, color = col_vec)
subplot(c(1,1)) #readjust the plotting area

Chitran1987/StatsChitran documentation built on Feb. 23, 2025, 8:30 p.m.