Quantile

Process Help

QUANTILE - perform quantile analysis on sample data

Process Name	Menu Path	Link to Command Table
QUANTILE	Sample Analysis ribbon \| Statistics Processes \| Calculate Quantiles	Click here

Introduction
Files, Fields and Parameters
Example
Notes

Introduction

This is a Superprocess and running it may have an effect on other Datamine files in the project. More...

Collectively quartiles, deciles, percentiles and other values obtained by equal subdivision of data are called quantiles. Quantiles give information about the shape of a distribution; in particular whether a distribution is skewed or not. Quantiles can be used for comparing two distributions.

The QUANTILE process carries out two types of analysis on a set of sample data:

Quantile Point Analysis: a table showing the grades corresponding to user defined percentile values.
Quantile Group Analysis: a set of tables showing the statistics of the grades lying between consecutive pairs of user defined percentile values

How to use

The first stage of the QUANTILE process is to apply the cutoff grade as specified by parameter CUTOFF. Any sample below the cutoff grade is removed from the analysis. Using the default cutoff grade of 0 in effect means that no cutoff is used.

The second stage is to apply a topcut grade. If the parameter TOPCUT has been set to 1, then any sample which is greater than parameter TOPGRADE will be replaced by a sample of TOPGRADE. If TOPCUT is set to 0, then the topcut will not be used.

The sample file is sorted in ascending order and divided into equal numbers of samples as defined by parameter QUANTIL1. For example, if there are 120 samples in total and QUANTIL1=10, then each subdivision or bin will include 12 samples. If the total number of samples does not divide equally by QUANTIL1, then some bins will contain one more sample than others. If QUANTIL1=8 then the first bin will contain the lowest grade 12.5% of samples, bin 2 will contain the samples between 12.5% and 25%, and so on, with the top bin, bin 8, contain the highest 12.5% of samples. The split using the QUANTIL1 parameter is called the primary subdivision.

The top bin can be further divided, as controlled by parameter QUANTIL2. For example if QUANTIL2=5, then 5 additional bins from 87.5% - 90%, 90% - 92.5%, etc will be calculated. This is the secondary subdivision. If you do not want a secondary subdivision then QUANTIL2 should be set to zero.

An example of a Quantile Point table, file QUANT_PT, is shown below. This has been created using QUANTIL1=10 and QUANTILE2=4. A set of results is shown for each zone if the KEY field has been selected:

The Group Analysis tables show statistics between each pair of Percentiles; 0% - 10%, 10% - 20%, and so on. The following statistics are calculated for each bin:

the number of samples
minimum grade
maximum grade
mean grade
metal content (the sum of the individual grade values)
% metal in bin (the metal content as a percentage of total metal)

Two Group Analysis output files can be created. The RESULTS file includes both the primary and secondary divisions and the PRIMARY file includes just the primary divisions. The Results table can be saved to a system text file if file PRINT is specified. The table will also be displayed in the Command Window. An example of the RESULTS file is shown below:

If a WEIGHT field has been specified then the mean grade is a weighted mean and the metal content is the sum of weight*grade. Also the values in the NSAMPLES (Number of Samples) column are not necessarily equal because the weights, not the number of samples, are equally distributed between the quantiles. The WEIGHT field applies to both the Point and Group analysis tables.

If a KEY field has been specified, then the quantile analysis is done separately for each value of the KEY field.

Files, Fields and Parameters

Input Files

Name	Description	I/O Status	Required	Type
IN	Input sample file	Input	Yes	Undefined

Output Files

Name	I/O Status	Required	Type	Description
QUANT_PT	Output	No	Table File	Output file containing the *VALUE value for each Quantile Point defined by parameters QUANTIL1 and QUANTIL2. Although optional one of the two files QUANT_PT or RESULTS must be selected.
RESULTS	Output	Yes	Table File	Output file containing quantile group information for primary and secondary subdivisions. Although optional one of the two files QUANT_PT or RESULTS must be selected.
PRIMARY	Output	No	Table File	Output file containing quantile group information for the primary subdivision only.
PRINT	Output	No	Table File	System print file, containing quantile group information. This is a copy of the contents of the RESULTS file, but to a system file. The extension .pri will be added automatically to the file name.

Fields

Name	Description	Source	Required	Type	Default
VALUE	Name of the field containing the grade to be analysed.	IN	Yes	Numeric	Undefined
KEY	Key field for grouping the data. A separate quantile analysis is carried out for each unique value of the key field.	IN	No	Numeric	Undefined
WEIGHT	Field containing the weight used when calculating quantile means and accumulating the grade values. For example if the input data is a desurveyed data file then LENGTH could be used. If no field is specified then all samples have an equal weight.	IN	No	Numeric	Undefined

Parameters

Name	Description	Required	Default	Range	Values
QUANTIL1	The primary quantile. The number of primary subdivisions or bins for grouping the samples. For example setting QUANTIL1=10 will divide the samples into deciles; QUANTIL1=4 will give quartiles.	No	10	2,+	Undefined
QUANTIL2	The secondary quantile. The top bin of the primary subdivision can be further split into equal groupings. For example if QUANTIL1=10 and QUANTIL2=4, then the top 10% will be split into 4 equal groups of 2.5%. If set to 0 then the top bin is not resplit.	No	0	0,+	Undefined
CUTOFF	Cutoff grade. Only samples greater than or equal to the cutoff grade are selected for analysis.	No	0	Undefined	Undefined
TOPCUT	Flag to specify whether or not a topcut grade should be applied: 0 = No topcut applied. 1 = Topcut applied at grade defined in TOPGRADE	No	0	0,1	0,1
TOPGRADE	Grade to be applied as a topcut, if TOPCUT is set to 1. Any value greater than TOPGRADE will be reset to equal to TOPGRADE.	No	Undefined	Undefined	Undefined
NDP	Maximum number of decimal places for field VALUE in output file QUANT_PT.	No	2	0,6	0,1,2,3,4,5,6

Example

!QUANTILE &IN(Data_Test_1), &QUANT_PT(Q_Points), &RESULTS(Q_Results),

&PRIMARY(Q_Primary), &PRINT(Q_Print), *VALUE(AG), *KEY(N_ZONE),

*WEIGHT(LENGTH), @QUANTIL1=10, @QUANTIL2=4, @CUTOFF=0,

@TOPCUT=0, @TOPGRADE=0, @NDP=2

QUANTILE - Sample quantile analysis.

... input validation

... checking files, fields and parameters

... key field N-ZONE is numeric

... applying cutoff, topcut, key and weights if required

... calculating primary quantiles

... calculating quantile points table

... calculating secondary quantiles

FORMAT TIME >11:36:36

________________________________________________________________________

Group Quantile Analysis for AG

-----------------------------------------------

Sample file: data_test_1 Key N-ZONE = 1.0000

Cutoff Grade = 0 Weighting Field LENGTH

------------------------------------------------------------------------

% Quantile No. of Mean Minimum Maximum Metal %

From To Samples Grade Grade Grade Content Metal

------------------------------------------------------------------------

0.0 10.0 14 20.890 6.680 39.750 177 0.8

10.0 20.0 11 48.067 41.900 57.200 445 1.9

20.0 30.0 12 67.154 57.750 76.180 543 2.4

30.0 40.0 13 90.739 76.230 107.800 855 3.7

40.0 50.0 12 141.987 112.140 167.100 1245 5.4

50.0 60.0 11 195.067 177.860 228.500 1736 7.5

60.0 70.0 8 258.704 235.400 279.390 2398 10.4

70.0 80.0 11 335.556 280.800 410.300 2981 13.0

80.0 90.0 8 531.775 422.100 649.280 4414 19.2

90.0 100.0 10 850.614 676.500 1509.190 8225 35.7

90.0 92.5 2 680.860 676.500 685.220 1362 5.9

92.5 95.0 3 688.160 685.800 695.250 1831 8.0

95.0 97.5 2 737.883 727.350 749.060 1491 6.5

97.5 100.0 3 1184.844 949.610 1509.190 3543 15.4

0.0 100.0 110 258.532 6.680 1509.190 23020 100.0

------------------------------------------------------------------------

OUTPUT FILES

The Quantile Points table is stored in Datamine file:

Q_Points 42 records

The Group Quantiles table is stored in Datamine file:

Q_Results 41 records

The Group Quantiles table has also been written to system file:

Q_Print.pri

The table for the primary quantile only is stored in Datamine file:

Q_Primary 30 records

________________________________________________________________________

... process complete.

Notes

The table for just one key value (N-ZONE = 1) has been included in the above output
The weighting field LENGTH has been used for the above example. This is different to the examples in the “How to use” section which did not include a weighting field. Therefore the results are different.