Logo ROOT  
Reference Guide
 
Loading...
Searching...
No Matches
TGraphQQ.cxx
Go to the documentation of this file.
1// @(#)root/graf:$Id$
2// Author: Anna Kreshuk 18/11/2005
3
4/*************************************************************************
5 * Copyright (C) 1995-2005, Rene Brun and Fons Rademakers. *
6 * All rights reserved. *
7 * *
8 * For the licensing terms see $ROOTSYS/LICENSE. *
9 * For the list of contributors see $ROOTSYS/README/CREDITS. *
10 *************************************************************************/
11
12#include "TGraphQQ.h"
13#include "TAxis.h"
14#include "TF1.h"
15#include "TMath.h"
16
18
19/** \class TGraphQQ
20\ingroup BasicGraphics
21
22This class allows to draw quantile-quantile plots
23
24Plots can be drawn for 2 datasets or for a dataset and a theoretical
25distribution function
26
27## 2 datasets:
28 Quantile-quantile plots are used to determine whether 2 samples come from
29 the same distribution.
30 A qq-plot draws the quantiles of one dataset against the quantile of the
31 the other. The quantiles of the dataset with fewer entries are on Y axis,
32 with more entries - on X axis.
33 A straight line, going through 0.25 and 0.75 quantiles is also plotted
34 for reference. It represents a robust linear fit, not sensitive to the
35 extremes of the datasets.
36 If the datasets come from the same distribution, points of the plot should
37 fall approximately on the 45 degrees line. If they have the same
38 distribution function, but location or scale different parameters,
39 they should still fall on the straight line, but not the 45 degrees one.
40 The greater their departure from the straight line, the more evidence there
41 is, that the datasets come from different distributions.
42 The advantage of qq-plot is that it not only shows that the underlying
43 distributions are different, but, unlike the analytical methods, it also
44 gives information on the nature of this difference: heavier tails,
45 different location/scale, different shape, etc.
46
47 Some examples of qqplots of 2 datasets:
48
49\image html graf_graphqq1.png
50
51## 1 dataset:
52 Quantile-quantile plots are used to determine if the dataset comes from the
53 specified theoretical distribution, such as normal.
54 A qq-plot draws quantiles of the dataset against quantiles of the specified
55 theoretical distribution.
56 (NOTE, that density, not CDF should be specified)
57 A straight line, going through 0.25 and 0.75 quantiles can also be plotted
58 for reference. It represents a robust linear fit, not sensitive to the
59 extremes of the dataset.
60 As in the 2 datasets case, departures from straight line indicate departures
61 from the specified distribution.
62
63 "The correlation coefficient associated with the linear fit to the data
64 in the probability plot (qq plot in our case) is a measure of the
65 goodness of the fit.
66 Estimates of the location and scale parameters of the distribution
67 are given by the intercept and slope. Probability plots can be generated
68 for several competing distributions to see which provides the best fit,
69 and the probability plot generating the highest correlation coefficient
70 is the best choice since it generates the straightest probability plot."
71
72 From "Engineering statistic handbook",
73
74 http://www.itl.nist.gov/div898/handbook/eda/section3/probplot.htm
75
76 Example of a qq-plot of a dataset from N(3, 2) distribution and
77 TMath::Gaus(0, 1) theoretical function. Fitting parameters
78 are estimates of the distribution mean and sigma.
79
80\image html graf_graphqq2.png
81
82References:
83
84http://www.itl.nist.gov/div898/handbook/eda/section3/qqplot.htm
85
86http://www.itl.nist.gov/div898/handbook/eda/section3/probplot.htm
87*/
88
89////////////////////////////////////////////////////////////////////////////////
90/// default constructor
91
93{
94 fF = 0;
95 fY0 = 0;
96 fNy0 = 0;
97 fXq1 = 0.;
98 fXq2 = 0.;
99 fYq1 = 0.;
100 fYq2 = 0.;
101
102}
103
104////////////////////////////////////////////////////////////////////////////////
105/// Creates a quantile-quantile plot of dataset x.
106/// Theoretical distribution function can be defined later by SetFunction method
107
109 : TGraph(n)
110{
111 fNy0 = 0;
112 fXq1 = 0.;
113 fXq2 = 0.;
114 fYq1 = 0.;
115 fYq2 = 0.;
116
117 Int_t *index = new Int_t[n];
118 TMath::Sort(n, x, index, kFALSE);
119 for (Int_t i=0; i<fNpoints; i++)
120 fY[i] = x[index[i]];
121 fF=0;
122 fY0=0;
123 delete [] index;
124}
125
126////////////////////////////////////////////////////////////////////////////////
127/// Creates a quantile-quantile plot of dataset x against function f
128
130 : TGraph(n)
131{
132 fNy0 = 0;
133
134 Int_t *index = new Int_t[n];
135 TMath::Sort(n, x, index, kFALSE);
136 for (Int_t i=0; i<fNpoints; i++)
137 fY[i] = x[index[i]];
138 delete [] index;
139 fF = f;
140 fY0=0;
142}
143
144////////////////////////////////////////////////////////////////////////////////
145/// Creates a quantile-quantile plot of dataset x against dataset y
146/// Parameters nx and ny are respective array sizes
147
149{
150 fNy0 = 0;
151 fXq1 = 0.;
152 fXq2 = 0.;
153 fYq1 = 0.;
154 fYq2 = 0.;
155 fF = 0;
156 fY0 = 0;
157
158 nx<=ny ? fNpoints=nx : fNpoints=ny;
159
160 if (!CtorAllocate()) return;
161
162 Int_t *index = new Int_t[TMath::Max(nx, ny)];
163 TMath::Sort(nx, x, index, kFALSE);
164 if (nx <=ny){
165 for (Int_t i=0; i<fNpoints; i++)
166 fY[i] = x[index[i]];
167 TMath::Sort(ny, y, index, kFALSE);
168 if (nx==ny){
169 for (Int_t i=0; i<fNpoints; i++)
170 fX[i] = y[index[i]];
171 fY0 = 0;
172 Quartiles();
173 } else {
174 fNy0 = ny;
175 fY0 = new Double_t[ny];
176 for (Int_t i=0; i<ny; i++)
177 fY0[i] = y[i];
179 }
180 } else {
181 fNy0 = nx;
182 fY0 = new Double_t[nx];
183 for (Int_t i=0; i<nx; i++)
184 fY0[i] = x[index[i]];
185 TMath::Sort(ny, y, index, kFALSE);
186 for (Int_t i=0; i<ny; i++)
187 fY[i] = y[index[i]];
189 }
190
191
192 delete [] index;
193}
194
195////////////////////////////////////////////////////////////////////////////////
196/// Destroys a TGraphQQ
197
199{
200 if (fY0)
201 delete [] fY0;
202 if (fF)
203 fF = 0;
204}
205
206////////////////////////////////////////////////////////////////////////////////
207/// Computes quantiles of theoretical distribution function
208
210{
211 if (!fF) return;
212 TString s = fF->GetTitle();
213 Double_t pk;
214 if (s.Contains("TMath::Gaus") || s.Contains("gaus")){
215 //use plotting positions optimal for normal distribution
216 for (Int_t k=1; k<=fNpoints; k++){
217 pk = (k-0.375)/(fNpoints+0.25);
218 fX[k-1]=TMath::NormQuantile(pk);
219 }
220 } else {
221 Double_t *prob = new Double_t[fNpoints];
222 if (fNpoints > 10){
223 for (Int_t k=1; k<=fNpoints; k++)
224 prob[k-1] = (k-0.5)/fNpoints;
225 } else {
226 for (Int_t k=1; k<=fNpoints; k++)
227 prob[k-1] = (k-0.375)/(fNpoints+0.25);
228 }
229 //fF->GetQuantiles(fNpoints, prob, fX);
230 fF->GetQuantiles(fNpoints, fX, prob);
231 delete [] prob;
232 }
233
234 Quartiles();
235}
236
237////////////////////////////////////////////////////////////////////////////////
238/// When sample sizes are not equal, computes quantiles of the bigger sample
239/// by linear interpolation
240
242{
243
244
245 if (!fY0) return;
246
247 Double_t pi, pfrac;
248 Int_t pint;
249 for (Int_t i=0; i<fNpoints-1; i++){
250 pi = (fNy0-1)*Double_t(i)/Double_t(fNpoints-1);
251 pint = TMath::FloorNint(pi);
252 pfrac = pi - pint;
253 fX[i] = (1-pfrac)*fY0[pint]+pfrac*fY0[pint+1];
254 }
255 fX[fNpoints-1]=fY0[fNy0-1];
256
257 Quartiles();
258}
259
260////////////////////////////////////////////////////////////////////////////////
261/// compute quartiles
262/// a quartile is a 25 per cent or 75 per cent quantile
263
265{
266 Double_t prob[]={0.25, 0.75};
267 Double_t x[2];
268 Double_t y[2];
269 TMath::Quantiles(fNpoints, 2, fY, y, prob, kTRUE);
270 if (fY0)
271 TMath::Quantiles(fNy0, 2, fY0, x, prob, kTRUE);
272 else if (fF) {
273 TString s = fF->GetTitle();
274 if (s.Contains("TMath::Gaus") || s.Contains("gaus")){
275 x[0] = TMath::NormQuantile(0.25);
276 x[1] = TMath::NormQuantile(0.75);
277 } else
278 fF->GetQuantiles(2, x, prob);
279 }
280 else
281 TMath::Quantiles(fNpoints, 2, fX, x, prob, kTRUE);
282
283 fXq1=x[0]; fXq2=x[1]; fYq1=y[0]; fYq2=y[1];
284}
285
286////////////////////////////////////////////////////////////////////////////////
287///Sets the theoretical distribution function (density!)
288///and computes its quantiles
289
291{
292 fF = f;
294}
#define f(i)
Definition RSha256.hxx:104
const Bool_t kFALSE
Definition RtypesCore.h:101
double Double_t
Definition RtypesCore.h:59
const Bool_t kTRUE
Definition RtypesCore.h:100
#define ClassImp(name)
Definition Rtypes.h:364
1-Dim function class
Definition TF1.h:213
virtual Int_t GetQuantiles(Int_t nprobSum, Double_t *q, const Double_t *probSum)
Compute Quantiles for density distribution of this function.
Definition TF1.cxx:1988
This class allows to draw quantile-quantile plots.
Definition TGraphQQ.h:18
void Quartiles()
compute quartiles a quartile is a 25 per cent or 75 per cent quantile
Definition TGraphQQ.cxx:264
TGraphQQ()
default constructor
Definition TGraphQQ.cxx:92
virtual ~TGraphQQ()
Destroys a TGraphQQ.
Definition TGraphQQ.cxx:198
TF1 * fF
theoretical density function, if specified
Definition TGraphQQ.h:26
Double_t fYq1
y1 coordinate of the interquartile line
Definition TGraphQQ.h:23
Double_t * fY0
! second dataset, if specified
Definition TGraphQQ.h:25
void SetFunction(TF1 *f)
Sets the theoretical distribution function (density!) and computes its quantiles.
Definition TGraphQQ.cxx:290
Double_t fXq2
x2 coordinate of the interquartile line
Definition TGraphQQ.h:22
Int_t fNy0
size of the fY0 dataset
Definition TGraphQQ.h:20
void MakeFunctionQuantiles()
Computes quantiles of theoretical distribution function.
Definition TGraphQQ.cxx:209
Double_t fXq1
x1 coordinate of the interquartile line
Definition TGraphQQ.h:21
Double_t fYq2
y2 coordinate of the interquartile line
Definition TGraphQQ.h:24
void MakeQuantiles()
When sample sizes are not equal, computes quantiles of the bigger sample by linear interpolation.
Definition TGraphQQ.cxx:241
A TGraph is an object made of two arrays X and Y with npoints each.
Definition TGraph.h:41
Int_t fNpoints
Number of points <= fMaxSize.
Definition TGraph.h:46
Double_t * fY
[fNpoints] array of Y points
Definition TGraph.h:48
Bool_t CtorAllocate()
In constructors set fNpoints than call this method.
Definition TGraph.cxx:743
Double_t * fX
[fNpoints] array of X points
Definition TGraph.h:47
virtual const char * GetTitle() const
Returns title of object.
Definition TNamed.h:48
Basic string class.
Definition TString.h:136
Bool_t Contains(const char *pat, ECaseCompare cmp=kExact) const
Definition TString.h:624
Double_t y[n]
Definition legend1.C:17
Double_t x[n]
Definition legend1.C:17
const Int_t n
Definition legend1.C:16
Short_t Max(Short_t a, Short_t b)
Definition TMathBase.h:208
Int_t FloorNint(Double_t x)
Definition TMath.h:657
Double_t NormQuantile(Double_t p)
Computes quantiles for standard normal distribution N(0, 1) at probability p.
Definition TMath.cxx:2416
void Sort(Index n, const Element *a, Index *index, Bool_t down=kTRUE)
Definition TMathBase.h:358
void Quantiles(Int_t n, Int_t nprob, Double_t *x, Double_t *quantiles, Double_t *prob, Bool_t isSorted=kTRUE, Int_t *index=0, Int_t type=7)
Computes sample quantiles, corresponding to the given probabilities.
Definition TMath.cxx:1183