Class: Statsample::StratifiedSample

Inherits:
Object
  • Object
show all
Defined in:
lib/statsample/multiset.rb

Class Method Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(ms, strata_sizes) ⇒ StratifiedSample

Returns a new instance of StratifiedSample.

Raises:

  • (TypeError) 


157
158
159
160
161
162
163
164
165
 
# File 'lib/statsample/multiset.rb', line 157

def initialize(ms,strata_sizes)
  raise TypeError,"ms should be a Multiset" unless ms.is_a? Statsample::Multiset
  @ms=ms
  raise ArgumentError,"You should put a strata size for each dataset" if strata_sizes.keys.sort!=ms.datasets_names
  @strata_sizes=strata_sizes
  @population_size=@strata_sizes.inject(0) {|a,x| a+x[1]}
  @strata_number=@ms.n_datasets
  @sample_size=@ms.datasets.inject(0) {|a,x| a+x[1].cases}
end

Class Method Details

.calculate_n_total(es) ⇒ Object 



75
76
77
 
# File 'lib/statsample/multiset.rb', line 75

def calculate_n_total(es)
  es.inject(0) {|a,h| a+h['N'] }
end

.mean(*vectors) ⇒ Object

mean for an array of vectors



53
54
55
56
57
58
59
60
 
# File 'lib/statsample/multiset.rb', line 53

def mean(*vectors)
  n_total=0
  means=vectors.inject(0){|a,v|
    n_total+=v.size
    a+v.sum
  }
  means.to_f/n_total
end

.proportion_sd_esd_wor(es) ⇒ Object 



152
153
154
 
# File 'lib/statsample/multiset.rb', line 152

def proportion_sd_esd_wor(es)
    Math::sqrt(proportion_variance_ksd_wor(es))
end

.proportion_sd_ksd_wor(es) ⇒ Object 



126
127
128
 
# File 'lib/statsample/multiset.rb', line 126

def proportion_sd_ksd_wor(es)
    Math::sqrt(proportion_variance_ksd_wor(es))
end

.proportion_sd_ksd_wr(es) ⇒ Object 



131
132
133
134
135
136
137
138
 
# File 'lib/statsample/multiset.rb', line 131

def proportion_sd_ksd_wr(es)
  n_total=calculate_n_total(es)
  sum=es.inject(0){|a,h|
    val= (h['N']**2 * h['p']*(1-h['p'])) / h['n'].to_f
    a+val
  }
  Math::sqrt(sum) * (1.0/n_total)
end

.proportion_variance_esd_wor(es) ⇒ Object 



143
144
145
146
147
148
149
150
151
 
# File 'lib/statsample/multiset.rb', line 143

def proportion_variance_esd_wor(es)
  n_total=n_total=calculate_n_total(es)
  
  sum=es.inject(0){|a,h|
    a=(h['N']**2 * (h['N']-h['n']) * h['p']*(1.0-h['p'])) / ((h['n']-1)*(h['N']-1))
    a+val
  }
  Math::sqrt(sum) * (1.0/n_total**2)
end

.proportion_variance_ksd_wor(es) ⇒ Object 



119
120
121
122
123
124
125
 
# File 'lib/statsample/multiset.rb', line 119

def proportion_variance_ksd_wor(es)
  n_total=calculate_n_total(es)
    es.inject(0){|a,h|
      val= (((h['N'].to_f / n_total)**2 * h['p']*(1-h['p'])) / (h['n'])) * (1- (h['n'].to_f / h['N']))
      a+val
    }
end

.proportion_variance_ksd_wr(es) ⇒ Object 



139
140
141
 
# File 'lib/statsample/multiset.rb', line 139

def proportion_variance_ksd_wr(es)
    proportion_variance_ksd_wor(es)**2
end

.standard_error_esd_wor(es) ⇒ Object 



103
104
105
 
# File 'lib/statsample/multiset.rb', line 103

def standard_error_esd_wor(es)
  Math::sqrt(variance_ksd_wor(es))
end

.standard_error_esd_wr(es) ⇒ Object 



115
116
117
 
# File 'lib/statsample/multiset.rb', line 115

def standard_error_esd_wr(es)
  Math::sqrt(variance_esd_wr(es))
end

.standard_error_ksd_wor(es) ⇒ Object 



87
88
89
 
# File 'lib/statsample/multiset.rb', line 87

def standard_error_ksd_wor(es)
  Math::sqrt(variance_ksd_wor(es))
end

.standard_error_ksd_wr(es) ⇒ Object 



62
63
64
65
66
67
68
69
 
# File 'lib/statsample/multiset.rb', line 62

def standard_error_ksd_wr(es)
  n_total=0
  sum=es.inject(0){|a,h|
      n_total+=h['N']
      a+((h['N']**2 * h['s']**2) / h['n'].to_f)
  }
  (1.to_f / n_total)*Math::sqrt(sum)
end

.variance_esd_wor(es) ⇒ Object 



93
94
95
96
97
98
99
100
 
# File 'lib/statsample/multiset.rb', line 93

def variance_esd_wor(es)
  n_total=calculate_n_total(es)
  sum=es.inject(0){|a,h|
    val=h['N']*(h['N']-h['n'])*(h['s']**2 / h['n'].to_f)
    a+val
  }
  (1.0/(n_total**2))*sum
end

.variance_esd_wr(es) ⇒ Object

Based on stattrek.com/Lesson6/STRAnalysis.aspx



107
108
109
110
111
112
113
114
 
# File 'lib/statsample/multiset.rb', line 107

def variance_esd_wr(es)
  n_total=calculate_n_total(es)
    sum=es.inject(0){|a,h|
      val= ((h['s']**2 * h['N']**2) / h['n'].to_f)
      a+val
    }
    (1.0/(n_total**2))*sum
end

.variance_ksd_wor(es) ⇒ Object

Source : Cochran (1972)



80
81
82
83
84
85
86
 
# File 'lib/statsample/multiset.rb', line 80

def variance_ksd_wor(es)
n_total=calculate_n_total(es)
es.inject(0){|a,h|
  val=((h['N'].to_f / n_total)**2) * (h['s']**2 / h['n'].to_f) * (1 - (h['n'].to_f / h['N']))
  a+val
}
end

.variance_ksd_wr(es) ⇒ Object 



72
73
74
 
# File 'lib/statsample/multiset.rb', line 72

def variance_ksd_wr(es)
  standard_error_ksd_wr(es)**2
end

Instance Method Details

#mean(field) ⇒ Object

Population mean based on strata



202
203
204
205
206
 
# File 'lib/statsample/multiset.rb', line 202

def mean(field)
  @ms.sum_field(field) {|s_name,vector|
  stratum_ponderation(s_name)*vector.mean
  }
end

#population_size ⇒ Object

Population size. Equal to sum of strata sizes Symbol: N<sub>h</sub>



172
173
174
 
# File 'lib/statsample/multiset.rb', line 172

def population_size
  @population_size
end

#proportion(field, v = 1) ⇒ Object

Population proportion based on strata



189
190
191
192
193
 
# File 'lib/statsample/multiset.rb', line 189

def proportion(field, v=1)
  @ms.sum_field(field) {|s_name,vector|
  stratum_ponderation(s_name)*vector.proportion(v)
  }
end

#proportion_sd_esd_wor(field, v = 1) ⇒ Object 



235
236
237
238
239
240
241
 
# File 'lib/statsample/multiset.rb', line 235

def proportion_sd_esd_wor(field,v=1)
  es=@ms.collect_vector(field) {|s_n, vector|
    {'N'=>@strata_sizes[s_n],'n'=>vector.size, 'p'=>vector.proportion(v)}
  }
  
  StratifiedSample.proportion_sd_esd_wor(es)
end

#proportion_standard_error(field, v = 1) ⇒ Object 



243
244
245
246
247
248
249
250
251
 
# File 'lib/statsample/multiset.rb', line 243

def proportion_standard_error(field,v=1)
  prop=proportion(field,v)
  sum=@ms.sum_field(field) {|s_name,vector|
    nh=vector.size
    s_size=@strata_sizes[s_name]
    (s_size**2 * (1-(nh/s_size)) * prop * (1-prop) / (nh -1 ))
  }
  (1.quo(@population_size)) * Math::sqrt(sum)
end

#sample_size ⇒ Object

Sample size. Equal to sum of sample of each stratum



176
177
178
 
# File 'lib/statsample/multiset.rb', line 176

def sample_size
  @sample_size
end

#standard_error_wor(field) ⇒ Object

Standard error with estimated population variance and without replacement. Source: Cochran (1972)



209
210
211
212
213
214
215
 
# File 'lib/statsample/multiset.rb', line 209

def standard_error_wor(field)
  es=@ms.collect_vector(field) {|s_n, vector|
    {'N'=>@strata_sizes[s_n],'n'=>vector.size, 's'=>vector.sds}
  }
  
  StratifiedSample.standard_error_esd_wor(es)
end

#standard_error_wor_2(field) ⇒ Object

Standard error with estimated population variance and without replacement. Source: stattrek.com/Lesson6/STRAnalysis.aspx



220
221
222
223
224
225
226
 
# File 'lib/statsample/multiset.rb', line 220

def standard_error_wor_2(field)
  sum=@ms.sum_field(field) {|s_name,vector|
    s_size=@strata_sizes[s_name]
  (s_size**2 * (1-(vector.size.to_f / s_size)) * vector.variance_sample / vector.size.to_f)
  }
  (1/@population_size.to_f)*Math::sqrt(sum)
end

#standard_error_wr(field) ⇒ Object 



228
229
230
231
232
233
234
 
# File 'lib/statsample/multiset.rb', line 228

def standard_error_wr(field)
  es=@ms.collect_vector(field) {|s_n, vector|
    {'N'=>@strata_sizes[s_n],'n'=>vector.size, 's'=>vector.sds}
  }
  
  StratifiedSample.standard_error_esd_wr(es)
end

#strata_number ⇒ Object

Number of strata



167
168
169
 
# File 'lib/statsample/multiset.rb', line 167

def strata_number
  @strata_number
end

#stratum_ponderation(h) ⇒ Object Also known as: wh

Stratum ponderation. Symbol: W<sub>h</sub>



196
197
198
 
# File 'lib/statsample/multiset.rb', line 196

def stratum_ponderation(h)
  @strata_sizes[h].to_f / @population_size
end

#stratum_size(h) ⇒ Object

Size of stratum x



180
181
182
 
# File 'lib/statsample/multiset.rb', line 180

def stratum_size(h)
  @strata_sizes[h]
end

#variance_pst(field, v = 1) ⇒ Object

Cochran(1971), p. 150



253
254
255
256
257
258
259
260
261
262
263
 
# File 'lib/statsample/multiset.rb', line 253

def variance_pst(field,v=1)
  sum=@ms.datasets.inject(0) {|a,da|
    stratum_name=da[0]
    ds=da[1]
    nh=ds.cases.to_f
    s_size=@strata_sizes[stratum_name]
    prop=ds[field].proportion(v)
    a + (((s_size**2 * (s_size-nh)) / (s_size-1))*(prop*(1-prop) / (nh-1)))
  }
  (1/@population_size.to_f ** 2)*sum
end

#vectors_by_field(field) ⇒ Object 



183
184
185
186
187
 
# File 'lib/statsample/multiset.rb', line 183

def vectors_by_field(field)
  @ms.datasets.collect{|k,ds|
    ds[field]
  }
end