quantiled#

Purpose#

Computes quantiles from data in a dataset, given specified probabilities.

Format#

y = quantiled(dataset, e, var)#

Parameters:

dataset (string) – dataset name, or NxM matrix of data.
e (Lx1 vector) – quantile levels or probabilities.
var (Kx1 vector or scalar zero, string array, or formula string.) –
If Kx1, character vector of labels selected for analysis, or numeric vector of column numbers in dataset of variables selected for analysis.

If var is scalar zero, all columns are selected.

If dataset is a matrix var cannot be a character vector.

If dataset includes variable names, then var could be a string array, e.g. "Height" $| "Weight" or formula string. e.g. "Height + Weight".

Returns:

y (LxK matrix) – quantiles.

Examples#

Use dataset name#

// Create file name with full path
file_name = getGAUSSHome("examples/fueleconomy.dat");

// Set up quantile levels
e = { .025, .5, .975 };

// Choose all variables in the dataset
var = 0;

// Compute quantiles
y = quantiled(file_name, e, var);

 print "medians";
 print y[2, .];
 print;
 print "95 percentiles";
 print y[1, .];
 print y[3, .];

produces:

medians
5000000        3.0000000

percentiles
5500000        1.4000000
0500000        6.2550000

Use .csv file and variable index#

// Create file name with full path
file_name = getGAUSSHome("examples/binary.csv");

// Set up quantile levels
e = { .025, .5, .975 };

// Set up variable index
var = 2|3;

// Compute quantiles
y = quantiled(file_name, e, var);

 print "medians";
 print y[2, .];
 print;
 print "95 percentiles";
 print y[1, .];
 print y[3, .];

After the above code:

medians
00000        3.3900000

percentiles
00000        2.6300000
00000        4.0000000

Use .xls file and formula string#

// Create file name with full path
file_name = getGAUSSHome("examples/nba_ht_wt.xls");

// Set up quantile levels
e = { .025, .5, .975 };

// Set up formula string
var = "Height + Weight" ;

// Compute quantiles
y = quantiled(file_name, e, var);
 print  "Height"$~"Weight";
 print  "medians";
 print y[2, .];
 print;
 print  "95 percentiles";
 print y[1, .];
 print y[3, .];

After the above code:

medians
00000        79.500000

percentiles
00000        72.000000
00000        84.000000

Remarks#

quantiled() will not succeed if N*minc(e) is less than 1, or N*maxc(e) is greater than $N - 1$. In other words, to produce a quantile for a level of .001, the input matrix must have more than 1000 rows.
The supported dataset types are CSV, XLS, XLSX, HDF5, FMT, DAT.
For HDF5 file, the dataset must include file schema and both file name and dataset name must be provided, e.g.
```
quantiled("h5://C:/gauss23/examples/testdata.h5/mydata", 0.5, 0).
```

Source#

quantile.src