1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

72

73

74

75

76

77

78

79

80

81

82

83

84

85

86

87

88

89

90

91

92

93

94

95

96

97

98

99

100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

121

122

123

124

125

126

127

128

129

130

131

132

133

134

135

136

137

138

139

140

141

142

143

144

145

146

147

148

149

150

151

152

153

154

155

156

157

158

159

160

161

162

163

164

165

166

167

168

169

170

171

172

173

174

175

176

177

178

179

180

181

182

183

184

185

186

187

188

189

190

191

192

193

194

195

196

197

198

199

200

201

202

203

204

205

206

207

208

209

210

211

212

213

214

215

216

217

218

219

220

221

222

223

224

225

226

227

228

229

230

231

232

233

234

235

236

237

238

239

240

241

242

243

244

245

246

247

248

249

250

251

252

253

254

255

256

257

258

from collections import namedtuple 

 

import numpy as np 

 

from . import distributions 

 

 

__all__ = ['_find_repeats', 'linregress', 'theilslopes'] 

 

LinregressResult = namedtuple('LinregressResult', ('slope', 'intercept', 

'rvalue', 'pvalue', 

'stderr')) 

 

def linregress(x, y=None): 

""" 

Calculate a linear least-squares regression for two sets of measurements. 

 

Parameters 

---------- 

x, y : array_like 

Two sets of measurements. Both arrays should have the same length. 

If only x is given (and y=None), then it must be a two-dimensional 

array where one dimension has length 2. The two sets of measurements 

are then found by splitting the array along the length-2 dimension. 

 

Returns 

------- 

slope : float 

slope of the regression line 

intercept : float 

intercept of the regression line 

rvalue : float 

correlation coefficient 

pvalue : float 

two-sided p-value for a hypothesis test whose null hypothesis is 

that the slope is zero, using Wald Test with t-distribution of 

the test statistic. 

stderr : float 

Standard error of the estimated gradient. 

 

See also 

-------- 

:func:`scipy.optimize.curve_fit` : Use non-linear 

least squares to fit a function to data. 

:func:`scipy.optimize.leastsq` : Minimize the sum of 

squares of a set of equations. 

 

Examples 

-------- 

>>> import matplotlib.pyplot as plt 

>>> from scipy import stats 

>>> np.random.seed(12345678) 

>>> x = np.random.random(10) 

>>> y = np.random.random(10) 

>>> slope, intercept, r_value, p_value, std_err = stats.linregress(x, y) 

 

To get coefficient of determination (r_squared) 

 

>>> print("r-squared:", r_value**2) 

r-squared: 0.08040226853902833 

 

Plot the data along with the fitted line 

 

>>> plt.plot(x, y, 'o', label='original data') 

>>> plt.plot(x, intercept + slope*x, 'r', label='fitted line') 

>>> plt.legend() 

>>> plt.show() 

 

""" 

TINY = 1.0e-20 

if y is None: # x is a (2, N) or (N, 2) shaped array_like 

x = np.asarray(x) 

if x.shape[0] == 2: 

x, y = x 

elif x.shape[1] == 2: 

x, y = x.T 

else: 

msg = ("If only `x` is given as input, it has to be of shape " 

"(2, N) or (N, 2), provided shape was %s" % str(x.shape)) 

raise ValueError(msg) 

else: 

x = np.asarray(x) 

y = np.asarray(y) 

 

if x.size == 0 or y.size == 0: 

raise ValueError("Inputs must not be empty.") 

 

n = len(x) 

xmean = np.mean(x, None) 

ymean = np.mean(y, None) 

 

# average sum of squares: 

ssxm, ssxym, ssyxm, ssym = np.cov(x, y, bias=1).flat 

r_num = ssxym 

r_den = np.sqrt(ssxm * ssym) 

if r_den == 0.0: 

r = 0.0 

else: 

r = r_num / r_den 

# test for numerical error propagation 

if r > 1.0: 

r = 1.0 

elif r < -1.0: 

r = -1.0 

 

df = n - 2 

slope = r_num / ssxm 

intercept = ymean - slope*xmean 

if n == 2: 

# handle case when only two points are passed in 

if y[0] == y[1]: 

prob = 1.0 

else: 

prob = 0.0 

sterrest = 0.0 

else: 

t = r * np.sqrt(df / ((1.0 - r + TINY)*(1.0 + r + TINY))) 

prob = 2 * distributions.t.sf(np.abs(t), df) 

sterrest = np.sqrt((1 - r**2) * ssym / ssxm / df) 

 

return LinregressResult(slope, intercept, r, prob, sterrest) 

 

 

def theilslopes(y, x=None, alpha=0.95): 

r""" 

Computes the Theil-Sen estimator for a set of points (x, y). 

 

`theilslopes` implements a method for robust linear regression. It 

computes the slope as the median of all slopes between paired values. 

 

Parameters 

---------- 

y : array_like 

Dependent variable. 

x : array_like or None, optional 

Independent variable. If None, use ``arange(len(y))`` instead. 

alpha : float, optional 

Confidence degree between 0 and 1. Default is 95% confidence. 

Note that `alpha` is symmetric around 0.5, i.e. both 0.1 and 0.9 are 

interpreted as "find the 90% confidence interval". 

 

Returns 

------- 

medslope : float 

Theil slope. 

medintercept : float 

Intercept of the Theil line, as ``median(y) - medslope*median(x)``. 

lo_slope : float 

Lower bound of the confidence interval on `medslope`. 

up_slope : float 

Upper bound of the confidence interval on `medslope`. 

 

Notes 

----- 

The implementation of `theilslopes` follows [1]_. The intercept is 

not defined in [1]_, and here it is defined as ``median(y) - 

medslope*median(x)``, which is given in [3]_. Other definitions of 

the intercept exist in the literature. A confidence interval for 

the intercept is not given as this question is not addressed in 

[1]_. 

 

References 

---------- 

.. [1] P.K. Sen, "Estimates of the regression coefficient based on Kendall's tau", 

J. Am. Stat. Assoc., Vol. 63, pp. 1379-1389, 1968. 

.. [2] H. Theil, "A rank-invariant method of linear and polynomial 

regression analysis I, II and III", Nederl. Akad. Wetensch., Proc. 

53:, pp. 386-392, pp. 521-525, pp. 1397-1412, 1950. 

.. [3] W.L. Conover, "Practical nonparametric statistics", 2nd ed., 

John Wiley and Sons, New York, pp. 493. 

 

Examples 

-------- 

>>> from scipy import stats 

>>> import matplotlib.pyplot as plt 

 

>>> x = np.linspace(-5, 5, num=150) 

>>> y = x + np.random.normal(size=x.size) 

>>> y[11:15] += 10 # add outliers 

>>> y[-5:] -= 7 

 

Compute the slope, intercept and 90% confidence interval. For comparison, 

also compute the least-squares fit with `linregress`: 

 

>>> res = stats.theilslopes(y, x, 0.90) 

>>> lsq_res = stats.linregress(x, y) 

 

Plot the results. The Theil-Sen regression line is shown in red, with the 

dashed red lines illustrating the confidence interval of the slope (note 

that the dashed red lines are not the confidence interval of the regression 

as the confidence interval of the intercept is not included). The green 

line shows the least-squares fit for comparison. 

 

>>> fig = plt.figure() 

>>> ax = fig.add_subplot(111) 

>>> ax.plot(x, y, 'b.') 

>>> ax.plot(x, res[1] + res[0] * x, 'r-') 

>>> ax.plot(x, res[1] + res[2] * x, 'r--') 

>>> ax.plot(x, res[1] + res[3] * x, 'r--') 

>>> ax.plot(x, lsq_res[1] + lsq_res[0] * x, 'g-') 

>>> plt.show() 

 

""" 

# We copy both x and y so we can use _find_repeats. 

y = np.array(y).flatten() 

if x is None: 

x = np.arange(len(y), dtype=float) 

else: 

x = np.array(x, dtype=float).flatten() 

if len(x) != len(y): 

raise ValueError("Incompatible lengths ! (%s<>%s)" % (len(y), len(x))) 

 

# Compute sorted slopes only when deltax > 0 

deltax = x[:, np.newaxis] - x 

deltay = y[:, np.newaxis] - y 

slopes = deltay[deltax > 0] / deltax[deltax > 0] 

slopes.sort() 

medslope = np.median(slopes) 

medinter = np.median(y) - medslope * np.median(x) 

# Now compute confidence intervals 

if alpha > 0.5: 

alpha = 1. - alpha 

 

z = distributions.norm.ppf(alpha / 2.) 

# This implements (2.6) from Sen (1968) 

_, nxreps = _find_repeats(x) 

_, nyreps = _find_repeats(y) 

nt = len(slopes) # N in Sen (1968) 

ny = len(y) # n in Sen (1968) 

# Equation 2.6 in Sen (1968): 

sigsq = 1/18. * (ny * (ny-1) * (2*ny+5) - 

sum(k * (k-1) * (2*k + 5) for k in nxreps) - 

sum(k * (k-1) * (2*k + 5) for k in nyreps)) 

# Find the confidence interval indices in `slopes` 

sigma = np.sqrt(sigsq) 

Ru = min(int(np.round((nt - z*sigma)/2.)), len(slopes)-1) 

Rl = max(int(np.round((nt + z*sigma)/2.)) - 1, 0) 

delta = slopes[[Rl, Ru]] 

return medslope, medinter, delta[0], delta[1] 

 

 

def _find_repeats(arr): 

# This function assumes it may clobber its input. 

if len(arr) == 0: 

return np.array(0, np.float64), np.array(0, np.intp) 

 

# XXX This cast was previously needed for the Fortran implementation, 

# should we ditch it? 

arr = np.asarray(arr, np.float64).ravel() 

arr.sort() 

 

# Taken from NumPy 1.9's np.unique. 

change = np.concatenate(([True], arr[1:] != arr[:-1])) 

unique = arr[change] 

change_idx = np.concatenate(np.nonzero(change) + ([arr.size],)) 

freq = np.diff(change_idx) 

atleast2 = freq > 1 

return unique[atleast2], freq[atleast2]