datafarming 1.3.0 → 2.1.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- checksums.yaml +4 -4
- data/LICENSE.md +1 -1
- data/README.md +10 -44
- data/datafarming.gemspec +12 -7
- data/exe/datafarming.rb +86 -0
- data/exe/generate_design.rb +304 -0
- data/lib/datafarming/freq_sets.rb +863 -208
- data/lib/datafarming/nolh_designs.rb +19105 -4483
- data/lib/datafarming/{factorial_generator.rb → res_v_seqs.rb} +31 -29
- data/lib/datafarming/scale_design.rb +29 -0
- data/lib/datafarming/screen_freq_sets.rb +143 -0
- data/lib/datafarming/version.rb +5 -0
- data/lib/datafarming.rb +8 -0
- metadata +48 -21
- data/exe/augment_design.rb +0 -66
- data/exe/scaled_fde.rb +0 -169
- data/exe/scaled_rf_cubed.rb +0 -173
- data/exe/stack_nolhs.rb +0 -174
data/exe/scaled_rf_cubed.rb
DELETED
|
@@ -1,173 +0,0 @@
|
|
|
1
|
-
#!/usr/bin/env ruby -w
|
|
2
|
-
|
|
3
|
-
require 'colorize'
|
|
4
|
-
String.disable_colorization false
|
|
5
|
-
|
|
6
|
-
require 'datafarming/error_handling'
|
|
7
|
-
require 'datafarming/factorial_generator'
|
|
8
|
-
|
|
9
|
-
def print_descriptor
|
|
10
|
-
description = [
|
|
11
|
-
"\nCreate a Resolution V fractional factorial design, either in standard +/-1",
|
|
12
|
-
'notation or scaled as per user-provided input. The design can optionally ',
|
|
13
|
-
'be supplemented with CCD center and star points. Output is written to',
|
|
14
|
-
'stdout'.blue + ' with formatting controlled by command-line options.'
|
|
15
|
-
]
|
|
16
|
-
STDERR.puts description
|
|
17
|
-
end
|
|
18
|
-
|
|
19
|
-
def print_directions # :nodoc:
|
|
20
|
-
prog = $PROGRAM_NAME.split(/\//)[-1]
|
|
21
|
-
help_msg = [
|
|
22
|
-
'', 'Syntax:', '',
|
|
23
|
-
" ruby #{prog} [Options] #factors | low/high-settings".yellow, '',
|
|
24
|
-
"The vertical bar '|' indicates valid alternatives.", '',
|
|
25
|
-
'Command-line args must specify either:',
|
|
26
|
-
' - the number of factors for the design; or',
|
|
27
|
-
' - low and hi values for all factors.', '',
|
|
28
|
-
"\nOptions:",
|
|
29
|
-
' --help | -h | -? | ?'.green,
|
|
30
|
-
"\tProduce this help message.",
|
|
31
|
-
' --design-only | -d'.green,
|
|
32
|
-
"\tprint the design without labeling",
|
|
33
|
-
' --csv | -c'.green,
|
|
34
|
-
"\tprint the design as comma separated",
|
|
35
|
-
' --center | -a'.green,
|
|
36
|
-
"\tadd global center pt to check linearity",
|
|
37
|
-
' --star | -s'.green,
|
|
38
|
-
"\taugment with Central Composite Design pts (includes --center)",
|
|
39
|
-
' --rotatable | -r'.green,
|
|
40
|
-
"\tscale a CCD design to be rotatable (includes --star)",
|
|
41
|
-
''
|
|
42
|
-
]
|
|
43
|
-
STDERR.puts help_msg
|
|
44
|
-
end
|
|
45
|
-
|
|
46
|
-
(print_descriptor && print_directions && exit) if ARGV.empty?
|
|
47
|
-
begin
|
|
48
|
-
labels = true
|
|
49
|
-
csv = false
|
|
50
|
-
scaled = false
|
|
51
|
-
center = false
|
|
52
|
-
star = false
|
|
53
|
-
rotatable = false
|
|
54
|
-
fmt_size = '5'
|
|
55
|
-
fmt_type = 'd'
|
|
56
|
-
|
|
57
|
-
while ARGV[0] =~ /^(-\D|\?)/
|
|
58
|
-
arg = ARGV.shift
|
|
59
|
-
case arg
|
|
60
|
-
when '-c', '--csv'
|
|
61
|
-
csv = true
|
|
62
|
-
when '-d', '--design-only'
|
|
63
|
-
labels = false
|
|
64
|
-
when '-a', '--center'
|
|
65
|
-
center = true
|
|
66
|
-
when '-s', '--star'
|
|
67
|
-
star = true
|
|
68
|
-
center = true
|
|
69
|
-
when '-r', '--rotatable'
|
|
70
|
-
rotatable = true
|
|
71
|
-
star = true
|
|
72
|
-
center = true
|
|
73
|
-
fmt_size = '11.8'
|
|
74
|
-
fmt_type = 'f'
|
|
75
|
-
when '-h', '--help', '-?', '?'
|
|
76
|
-
print_descriptor
|
|
77
|
-
print_directions
|
|
78
|
-
exit
|
|
79
|
-
else
|
|
80
|
-
raise "Unknown option #{arg}"
|
|
81
|
-
end
|
|
82
|
-
end
|
|
83
|
-
|
|
84
|
-
if ARGV.empty?
|
|
85
|
-
raise 'Must specify a design size'
|
|
86
|
-
elsif ARGV.length == 1
|
|
87
|
-
n = ARGV.shift.to_i
|
|
88
|
-
elsif (ARGV.length & 1).nonzero? # Odd number of args remaining!
|
|
89
|
-
raise 'Number of levels is odd: low/hi values must occur in pairs'
|
|
90
|
-
else
|
|
91
|
-
n = ARGV.length / 2
|
|
92
|
-
scaled = true
|
|
93
|
-
unless rotatable
|
|
94
|
-
fmt_type = 's'
|
|
95
|
-
fmt_size = [ARGV.map(&:length).max + 1, fmt_size.to_i].max
|
|
96
|
-
end
|
|
97
|
-
end
|
|
98
|
-
|
|
99
|
-
my_design = make_design n
|
|
100
|
-
if rotatable && star
|
|
101
|
-
inverse_len = 1.0 / Math.sqrt(n)
|
|
102
|
-
my_design.each_with_index do |row, i|
|
|
103
|
-
my_design[i] = row.map { |elt| elt * inverse_len }
|
|
104
|
-
end
|
|
105
|
-
end
|
|
106
|
-
my_design += [Array.new(n, 0)] if center
|
|
107
|
-
|
|
108
|
-
if star
|
|
109
|
-
star_design = Array.new(2 * n)
|
|
110
|
-
star_design.each_index { |row| star_design[row] = Array.new(n, 0) }
|
|
111
|
-
n.times do |col|
|
|
112
|
-
row = 2 * col
|
|
113
|
-
star_design[row][col] = -1
|
|
114
|
-
star_design[row + 1][col] = 1
|
|
115
|
-
end
|
|
116
|
-
my_design += star_design
|
|
117
|
-
end
|
|
118
|
-
|
|
119
|
-
if scaled
|
|
120
|
-
temp = my_design.transpose
|
|
121
|
-
temp.each_with_index do |factor, idx|
|
|
122
|
-
factor.map! do |value|
|
|
123
|
-
case value
|
|
124
|
-
when -1
|
|
125
|
-
ARGV[2 * idx + 1]
|
|
126
|
-
when 1
|
|
127
|
-
ARGV[2 * idx]
|
|
128
|
-
when 0
|
|
129
|
-
begin
|
|
130
|
-
0.5 * (Float(ARGV[2 * idx + 1]) + Float(ARGV[2 * idx]))
|
|
131
|
-
rescue
|
|
132
|
-
raise 'All factors must be numeric for augmented designs.'
|
|
133
|
-
end
|
|
134
|
-
else
|
|
135
|
-
begin
|
|
136
|
-
lo = Float(ARGV[2 * idx])
|
|
137
|
-
hi = Float(ARGV[2 * idx + 1])
|
|
138
|
-
range = hi - lo
|
|
139
|
-
0.5 * (hi + lo + value * range)
|
|
140
|
-
rescue
|
|
141
|
-
raise 'All factors must be numeric for augmented designs.'
|
|
142
|
-
end
|
|
143
|
-
end
|
|
144
|
-
end
|
|
145
|
-
end
|
|
146
|
-
my_design = temp.transpose
|
|
147
|
-
end
|
|
148
|
-
|
|
149
|
-
if labels
|
|
150
|
-
if csv
|
|
151
|
-
print ' DP#'
|
|
152
|
-
n.times { |i| print ",X#{i + 1}" }
|
|
153
|
-
else
|
|
154
|
-
print ' DP#'
|
|
155
|
-
n.times { |i| printf " %#{fmt_size}s", "X#{i + 1}" }
|
|
156
|
-
end
|
|
157
|
-
puts
|
|
158
|
-
end
|
|
159
|
-
|
|
160
|
-
my_design.each_with_index do |design_point, row_number|
|
|
161
|
-
if csv
|
|
162
|
-
printf('%d,', row_number + 1) if labels
|
|
163
|
-
puts design_point.join(',')
|
|
164
|
-
else
|
|
165
|
-
printf('%5d', row_number + 1) if labels
|
|
166
|
-
design_point.each { |value| printf " %#{fmt_size}#{fmt_type}", value }
|
|
167
|
-
puts
|
|
168
|
-
end
|
|
169
|
-
end
|
|
170
|
-
rescue StandardError => e
|
|
171
|
-
STDERR.print "\n", e.message, "\n"
|
|
172
|
-
print_directions
|
|
173
|
-
end
|
data/exe/stack_nolhs.rb
DELETED
|
@@ -1,174 +0,0 @@
|
|
|
1
|
-
#!/usr/bin/env ruby -w
|
|
2
|
-
|
|
3
|
-
require 'colorize'
|
|
4
|
-
String.disable_colorization false
|
|
5
|
-
|
|
6
|
-
require 'datafarming/error_handling'
|
|
7
|
-
require 'datafarming/nolh_designs'
|
|
8
|
-
|
|
9
|
-
help_msg = [
|
|
10
|
-
'Generate scaled Latin hypercube designs with shifting and stacking. ',
|
|
11
|
-
'Results are a white-space delimited NOLH design written to ' +
|
|
12
|
-
'stdout'.light_blue + '.', '',
|
|
13
|
-
'Syntax:',
|
|
14
|
-
"\n\t#{$PROGRAM_NAME.split(%r{/|\\})[-1]} [--help]".yellow +
|
|
15
|
-
" [--stack #] [--levels #] [-e] [file_name]\n".yellow,
|
|
16
|
-
"Arguments in square brackets are optional. A vertical bar '|'",
|
|
17
|
-
'indicates valid alternatives for invoking the option. Prefix',
|
|
18
|
-
'the command with "' + 'ruby'.yellow +
|
|
19
|
-
'" if it is not on your PATH.', '',
|
|
20
|
-
' --help | -h | -? | ?'.green,
|
|
21
|
-
"\tProduce this help message. Supersedes any other choices.",
|
|
22
|
-
' --stack # | -s #'.green,
|
|
23
|
-
"\t# specifies the number of stackings. A value of 1 means print the",
|
|
24
|
-
"\tbase design. If this option is not specified the number of stackings",
|
|
25
|
-
"\tdefaults to the number of columns in the design. The specified value",
|
|
26
|
-
"\tcannot exceed the number of columns in the design being used.",
|
|
27
|
-
' --levels # | -l #'.green,
|
|
28
|
-
"\t# specifies the desired number of levels in the NOLH (17, 33, 65, 129,",
|
|
29
|
-
"\t257, or 512). Defaults to the smallest design which can accommodate the",
|
|
30
|
-
"\tnumber of factors if this option is not specified.",
|
|
31
|
-
' --excel-style-input | -e'.green,
|
|
32
|
-
"\tSpecify factor ranges and decimals as in the NOLH spreadsheet, i.e.,",
|
|
33
|
-
"\tthe first line is the set of minimum range values for each factor;",
|
|
34
|
-
"\tthe second line is maximum range values; and the third is the number",
|
|
35
|
-
"\tof decimal places to use for the range scaling. Without this option,",
|
|
36
|
-
"\tthe default input format is one line per factor, comprised of the min,",
|
|
37
|
-
"\tmax, and number of decimal places separated by commas or whitespace.",
|
|
38
|
-
' file_name'.green,
|
|
39
|
-
"\tThe name of a file containing the factor specifications. If no",
|
|
40
|
-
"\tfilename is given, the user can enter the values interactively in",
|
|
41
|
-
"\tthe desired form or use file redirection with '<'.", '',
|
|
42
|
-
'Options may be given in any order, but must come before the file name',
|
|
43
|
-
'if one is provided.'
|
|
44
|
-
]
|
|
45
|
-
|
|
46
|
-
# Scaler objects will rescale a Latin Hypercube design from standard units
|
|
47
|
-
# to a range as specified by min, max, and num_decimals
|
|
48
|
-
class Scaler
|
|
49
|
-
def initialize(min, max, num_decimals, lh_max = 17)
|
|
50
|
-
@min = min
|
|
51
|
-
@range = (max - min) / (lh_max - 1).to_r
|
|
52
|
-
@scale_factor = 10.to_r**num_decimals
|
|
53
|
-
end
|
|
54
|
-
|
|
55
|
-
def scale(value)
|
|
56
|
-
new_value = @min + @range * (value.to_r - 1.to_r)
|
|
57
|
-
if @scale_factor == 1
|
|
58
|
-
new_value.round
|
|
59
|
-
else
|
|
60
|
-
((@scale_factor * new_value).round / @scale_factor).to_f
|
|
61
|
-
end
|
|
62
|
-
end
|
|
63
|
-
end
|
|
64
|
-
|
|
65
|
-
excel_style_inputs = false
|
|
66
|
-
while ARGV[0] && (ARGV[0][0] == '-' || ARGV[0][0] == 45 || ARGV[0][0] == '?')
|
|
67
|
-
current_value = ARGV.shift
|
|
68
|
-
case current_value
|
|
69
|
-
when '--stack', '-s'
|
|
70
|
-
num_stackings = ARGV.shift.to_i
|
|
71
|
-
when '--levels', '-l'
|
|
72
|
-
lh_levels = ARGV.shift.to_i
|
|
73
|
-
unless NOLH::DESIGN_TABLE.keys.include?(lh_levels)
|
|
74
|
-
ErrorHandling.clean_abort [
|
|
75
|
-
"Invalid number of levels for Latin hypercube: #{lh_levels}".red,
|
|
76
|
-
'Use 17, 33, 65, 129, 257, or 512.'.yellow
|
|
77
|
-
]
|
|
78
|
-
end
|
|
79
|
-
when '--excel-style-input', '-e'
|
|
80
|
-
excel_style_inputs = true
|
|
81
|
-
when '--help', '-h', '-help', '-?', '?'
|
|
82
|
-
ErrorHandling.clean_abort help_msg
|
|
83
|
-
else
|
|
84
|
-
ErrorHandling.message ['Unknown argument: '.red + current_value.yellow]
|
|
85
|
-
ErrorHandling.clean_abort help_msg
|
|
86
|
-
end
|
|
87
|
-
end
|
|
88
|
-
|
|
89
|
-
begin
|
|
90
|
-
if excel_style_inputs
|
|
91
|
-
if ARGV.empty?
|
|
92
|
-
STDERR.puts 'Enter one line of min values, one of max values,'.green +
|
|
93
|
-
' and one of #decimals.'.green
|
|
94
|
-
end
|
|
95
|
-
min_values = ARGF.gets.strip.split(/\s*[,;:]\s*|\s+/).map(&:to_f)
|
|
96
|
-
max_values = ARGF.gets.strip.split(/\s*[,;:]\s*|\s+/).map(&:to_f)
|
|
97
|
-
decimals = ARGF.gets.strip.split(/\s*[,;:]\s*|\s+/).map(&:to_i)
|
|
98
|
-
else
|
|
99
|
-
if ARGV.empty?
|
|
100
|
-
STDERR.puts 'To terminate input enter '.green + 'ctrl-d'.cyan +
|
|
101
|
-
' (Mac/Unix/Linux)'.green + ' or '.green + 'ctrl-z'.cyan +
|
|
102
|
-
' (Windows).'.green
|
|
103
|
-
STDERR.puts 'Enter ranges for each factor on a separate line.'.green
|
|
104
|
-
STDERR.puts "\nMIN\tMAX\t#DIGITS".cyan
|
|
105
|
-
end
|
|
106
|
-
min_values = []
|
|
107
|
-
max_values = []
|
|
108
|
-
decimals = []
|
|
109
|
-
while line = ARGF.gets
|
|
110
|
-
values = line.strip.split(/\s*[,;:]\s*|\s+/)
|
|
111
|
-
min_values << values.shift.to_f
|
|
112
|
-
max_values << values.shift.to_f
|
|
113
|
-
decimals << values.shift.to_i
|
|
114
|
-
end
|
|
115
|
-
end
|
|
116
|
-
rescue StandardError => e
|
|
117
|
-
ErrorHandling.message [e.message.red]
|
|
118
|
-
ErrorHandling.clean_abort help_msg
|
|
119
|
-
end
|
|
120
|
-
|
|
121
|
-
n = min_values.size
|
|
122
|
-
if max_values.size != n || decimals.size != n
|
|
123
|
-
ErrorHandling.message ['Unequal counts for min, max, and decimals'.red]
|
|
124
|
-
ErrorHandling.clean_abort help_msg
|
|
125
|
-
end
|
|
126
|
-
minimal_size = case min_values.size
|
|
127
|
-
when 1..7
|
|
128
|
-
17
|
|
129
|
-
when 8..11
|
|
130
|
-
33
|
|
131
|
-
when 12..16
|
|
132
|
-
65
|
|
133
|
-
when 17..22
|
|
134
|
-
129
|
|
135
|
-
when 23..29
|
|
136
|
-
257
|
|
137
|
-
when 30..100
|
|
138
|
-
512
|
|
139
|
-
else
|
|
140
|
-
ErrorHandling.message ['invalid number of factors'.red]
|
|
141
|
-
ErrorHandling.clean_abort help_msg
|
|
142
|
-
end
|
|
143
|
-
|
|
144
|
-
lh_levels ||= minimal_size
|
|
145
|
-
|
|
146
|
-
if lh_levels < minimal_size
|
|
147
|
-
ErrorHandling.clean_abort [
|
|
148
|
-
"Latin hypercube with #{lh_levels} levels is too small for #{n} factors.".red
|
|
149
|
-
]
|
|
150
|
-
end
|
|
151
|
-
|
|
152
|
-
factor = Array.new(n) do |i|
|
|
153
|
-
Scaler.new(min_values[i], max_values[i], decimals[i], lh_levels)
|
|
154
|
-
end
|
|
155
|
-
|
|
156
|
-
design = NOLH::DESIGN_TABLE[lh_levels]
|
|
157
|
-
|
|
158
|
-
num_columns = design[0].length
|
|
159
|
-
num_stackings ||= num_columns
|
|
160
|
-
if num_stackings > num_columns
|
|
161
|
-
ErrorHandling.clean_abort [
|
|
162
|
-
'Requested stacking exceeds number of columns in latin hypercube '.red +
|
|
163
|
-
"(#{num_columns})".red
|
|
164
|
-
]
|
|
165
|
-
end
|
|
166
|
-
|
|
167
|
-
mid_range = lh_levels / 2
|
|
168
|
-
num_stackings.times do |stack_num|
|
|
169
|
-
design.each_with_index do |dp, i|
|
|
170
|
-
scaled_dp = dp.slice(0, n).map.with_index { |x, k| factor[k].scale(x) }
|
|
171
|
-
puts scaled_dp.join "\t" unless stack_num > 0 && i == mid_range && lh_levels < 512
|
|
172
|
-
design[i] = dp.rotate
|
|
173
|
-
end
|
|
174
|
-
end
|