davidrichards-data_frame 0.0.15 → 0.0.17
Sign up to get free protection for your applications and to get access to all the features.
- data/README.rdoc +15 -0
- data/VERSION.yml +1 -1
- data/lib/data_frame.rb +3 -302
- data/lib/data_frame/data_frame.rb +301 -0
- data/lib/data_frame/model.rb +22 -0
- data/lib/data_frame/parameter_capture.rb +50 -0
- data/spec/data_frame/data_frame_spec.rb +341 -0
- data/spec/data_frame/model_spec.rb +36 -0
- data/spec/data_frame/parameter_capture_spec.rb +32 -0
- data/spec/data_frame_spec.rb +14 -326
- metadata +9 -4
data/README.rdoc
CHANGED
@@ -75,6 +75,21 @@ A lot of the work in the data frame is to transform the actual table. You may n
|
|
75
75
|
|
76
76
|
|
77
77
|
Note: most of these transformations are not optimized. I'll work with things for a while before I try to optimize this library. However, I should say that I've used some fairly large data sets (thousands of rows) and have been fine with things so far.
|
78
|
+
|
79
|
+
== Models
|
80
|
+
|
81
|
+
Data Frame can now create sub-models:
|
82
|
+
|
83
|
+
>> df = DataFrame.from_csv('http://archive.ics.uci.edu/ml/machine-learning-databases/forest-fires/forestfires.csv')
|
84
|
+
=> DataFrame rows: 517 labels: [:x, :y, :month, :day, :ffmc, :dmc, :dc, :isi, :temp, :rh, :wind, :rain, :area]
|
85
|
+
>> df.model(:weekend) do |m|
|
86
|
+
?> m.day %w(sat sun)
|
87
|
+
>> end
|
88
|
+
=> DataFrame rows: 179 labels: [:x, :y, :month, :day, :ffmc, :dmc, :dc, :isi, :temp, :rh, :wind, :rain, :area]
|
89
|
+
>> df.models.weekend.day.uniq
|
90
|
+
=> ["sat", "sun"]
|
91
|
+
>> df.models
|
92
|
+
=> #<OpenStruct weekend=DataFrame rows: 179 labels: [:x, :y, :month, :day, :ffmc, :dmc, :dc, :isi, :temp, :rh, :wind, :rain, :area]>
|
78
93
|
|
79
94
|
==Installation
|
80
95
|
|
data/VERSION.yml
CHANGED
data/lib/data_frame.rb
CHANGED
@@ -19,305 +19,6 @@ $:.unshift(File.dirname(__FILE__))
|
|
19
19
|
|
20
20
|
require 'data_frame/callback_array'
|
21
21
|
require 'data_frame/transposable_array'
|
22
|
-
|
23
|
-
|
24
|
-
|
25
|
-
# transpose the matrix and store the transposed matrix until the object
|
26
|
-
# is tainted.
|
27
|
-
class DataFrame
|
28
|
-
|
29
|
-
class << self
|
30
|
-
|
31
|
-
# This is the neatest part of this neat gem.
|
32
|
-
# DataFrame.from_csv can be called in a lot of ways:
|
33
|
-
# DataFrame.from_csv(csv_contents)
|
34
|
-
# DataFrame.from_csv(filename)
|
35
|
-
# DataFrame.from_csv(url)
|
36
|
-
# If you need to define converters for FasterCSV, do it before calling
|
37
|
-
# this method:
|
38
|
-
# FasterCSV::Converters[:special] = lambda{|f| f == 'foo' ? 'bar' : 'foo'}
|
39
|
-
# DataFrame.from_csv('http://example.com/my_special_url.csv', :converters => :special)
|
40
|
-
# This returns bar where 'foo' was found and 'foo' everywhere else.
|
41
|
-
def from_csv(obj, opts={})
|
42
|
-
labels, table = infer_csv_contents(obj, opts)
|
43
|
-
return nil unless labels and table
|
44
|
-
df = new(*labels)
|
45
|
-
df.import(table)
|
46
|
-
df
|
47
|
-
end
|
48
|
-
|
49
|
-
protected
|
50
|
-
def infer_csv_contents(obj, opts={})
|
51
|
-
contents = File.read(obj) if File.exist?(obj)
|
52
|
-
begin
|
53
|
-
open(obj) {|f| contents = f.read} unless contents
|
54
|
-
rescue
|
55
|
-
nil
|
56
|
-
end
|
57
|
-
contents ||= obj if obj.is_a?(String)
|
58
|
-
return nil unless contents
|
59
|
-
table = FCSV.parse(contents, default_csv_opts.merge(opts))
|
60
|
-
labels = table.shift
|
61
|
-
while table.last.empty?
|
62
|
-
table.pop
|
63
|
-
end
|
64
|
-
[labels, table]
|
65
|
-
end
|
66
|
-
|
67
|
-
def default_csv_opts; {:converters => :all}; end
|
68
|
-
end
|
69
|
-
|
70
|
-
# Loads a batch of rows. Expects an array of arrays, else you don't
|
71
|
-
# know what you have.
|
72
|
-
def import(rows)
|
73
|
-
rows.each do |row|
|
74
|
-
self.add_item(row)
|
75
|
-
end
|
76
|
-
end
|
77
|
-
|
78
|
-
def inspect
|
79
|
-
"DataFrame rows: #{self.rows.size} labels: #{self.labels.inspect}"
|
80
|
-
end
|
81
|
-
|
82
|
-
# The labels of the data items
|
83
|
-
attr_reader :labels
|
84
|
-
alias :variables :labels
|
85
|
-
|
86
|
-
# The items stored in the frame
|
87
|
-
attr_reader :items
|
88
|
-
|
89
|
-
def initialize(*labels)
|
90
|
-
@labels = labels.map {|e| e.to_underscore_sym }
|
91
|
-
@items = TransposableArray.new
|
92
|
-
end
|
93
|
-
|
94
|
-
def add_item(item)
|
95
|
-
self.items << item
|
96
|
-
end
|
97
|
-
alias :add :add_item
|
98
|
-
|
99
|
-
def row_labels
|
100
|
-
@row_labels ||= []
|
101
|
-
end
|
102
|
-
|
103
|
-
def row_labels=(ary)
|
104
|
-
raise ArgumentError, "Row labels must be an array" unless ary.is_a?(Array)
|
105
|
-
@row_labels = ary
|
106
|
-
end
|
107
|
-
|
108
|
-
def render_column(sym)
|
109
|
-
i = @labels.index(sym)
|
110
|
-
return nil unless i
|
111
|
-
@items.transpose[i]
|
112
|
-
end
|
113
|
-
|
114
|
-
# The rows as an array of arrays, an alias for items.
|
115
|
-
alias :rows :items
|
116
|
-
|
117
|
-
# The columns as a Dictionary or Hash
|
118
|
-
# This is cached, call columns(true) to reset the cache.
|
119
|
-
def columns(reset=false)
|
120
|
-
@columns = nil if reset
|
121
|
-
return @columns if @columns
|
122
|
-
|
123
|
-
container = defined?(Dictionary) ? Dictionary.new : Hash.new
|
124
|
-
i = 0
|
125
|
-
|
126
|
-
@columns = @items.transpose.inject(container) do |cont, col|
|
127
|
-
cont[@labels[i]] = col
|
128
|
-
i += 1
|
129
|
-
cont
|
130
|
-
end
|
131
|
-
end
|
132
|
-
alias :to_hash :columns
|
133
|
-
alias :to_dictionary :columns
|
134
|
-
|
135
|
-
def render_row(sym)
|
136
|
-
i = self.row_labels.index(sym)
|
137
|
-
return nil unless i
|
138
|
-
@items[i]
|
139
|
-
end
|
140
|
-
|
141
|
-
def method_missing(sym, *args, &block)
|
142
|
-
if self.labels.include?(sym)
|
143
|
-
render_column(sym)
|
144
|
-
elsif self.row_labels.include?(sym)
|
145
|
-
render_row(sym)
|
146
|
-
elsif @items.respond_to?(sym)
|
147
|
-
@items.send(sym, *args, &block)
|
148
|
-
else
|
149
|
-
super
|
150
|
-
end
|
151
|
-
end
|
152
|
-
|
153
|
-
def drop!(*labels)
|
154
|
-
labels.each do |label|
|
155
|
-
drop_one!(label)
|
156
|
-
end
|
157
|
-
self
|
158
|
-
end
|
159
|
-
|
160
|
-
def drop_one!(label)
|
161
|
-
i = self.labels.index(label)
|
162
|
-
return nil unless i
|
163
|
-
self.items.each do |item|
|
164
|
-
item.delete_at(i)
|
165
|
-
end
|
166
|
-
self.labels.delete_at(i)
|
167
|
-
self
|
168
|
-
end
|
169
|
-
protected :drop_one!
|
170
|
-
|
171
|
-
def replace!(column, values=nil, &block)
|
172
|
-
column = validate_column(column)
|
173
|
-
if not values
|
174
|
-
values = self.send(column)
|
175
|
-
values.map! {|e| block.call(e)}
|
176
|
-
end
|
177
|
-
replace_column(column, values)
|
178
|
-
self
|
179
|
-
end
|
180
|
-
|
181
|
-
def replace_column(column, values)
|
182
|
-
column = validate_column(column)
|
183
|
-
index = self.labels.index(column)
|
184
|
-
list = []
|
185
|
-
self.items.each_with_index do |item, i|
|
186
|
-
consolidated = item
|
187
|
-
consolidated[index] = values[i]
|
188
|
-
list << consolidated
|
189
|
-
end
|
190
|
-
@items = list.dup
|
191
|
-
end
|
192
|
-
protected :replace_column
|
193
|
-
|
194
|
-
def validate_column(column)
|
195
|
-
column = column.to_sym
|
196
|
-
raise ArgumentError, "Must provide the name of an existing column. Provided #{column.inspect}, needed to provide one of #{self.labels.inspect}" unless self.labels.include?(column)
|
197
|
-
column
|
198
|
-
end
|
199
|
-
protected :validate_column
|
200
|
-
|
201
|
-
# Takes a block to evaluate on each row. The row can be converted into
|
202
|
-
# an OpenStruct or a Hash for easier filter methods. Note, don't try this
|
203
|
-
# with a hash or open struct unless you have facets available.
|
204
|
-
def filter!(as=Array, &block)
|
205
|
-
as = infer_class(as)
|
206
|
-
items = []
|
207
|
-
self.items.each do |row|
|
208
|
-
value = block.call(cast_row(row, as))
|
209
|
-
items << row if value
|
210
|
-
end
|
211
|
-
@items = items.dup
|
212
|
-
self
|
213
|
-
end
|
214
|
-
|
215
|
-
def filter(as=Array, &block)
|
216
|
-
new_data_frame = self.clone
|
217
|
-
new_data_frame.filter!(as, &block)
|
218
|
-
end
|
219
|
-
|
220
|
-
def infer_class(obj)
|
221
|
-
obj = obj.to_s.classify.constantize if obj.is_a?(Symbol)
|
222
|
-
obj = obj.classify.constantize if obj.is_a?(String)
|
223
|
-
obj
|
224
|
-
end
|
225
|
-
protected :infer_class
|
226
|
-
|
227
|
-
def cast_row(row, as)
|
228
|
-
if as == Hash
|
229
|
-
obj = {}
|
230
|
-
self.labels.each_with_index do |label, i|
|
231
|
-
obj[label] = row[i]
|
232
|
-
end
|
233
|
-
obj
|
234
|
-
elsif as == OpenStruct
|
235
|
-
obj = OpenStruct.new
|
236
|
-
self.labels.each_with_index do |label, i|
|
237
|
-
obj.table[label] = row[i]
|
238
|
-
end
|
239
|
-
obj
|
240
|
-
elsif as == Array
|
241
|
-
row
|
242
|
-
else
|
243
|
-
as.new(*row)
|
244
|
-
end
|
245
|
-
end
|
246
|
-
protected :cast_row
|
247
|
-
|
248
|
-
# Creates a new data frame, only with the specified columns.
|
249
|
-
def subset_from_columns(*cols)
|
250
|
-
new_labels = self.labels.inject([]) do |list, label|
|
251
|
-
list << label if cols.include?(label)
|
252
|
-
list
|
253
|
-
end
|
254
|
-
new_data_frame = DataFrame.new(*self.labels)
|
255
|
-
new_data_frame.import(self.items)
|
256
|
-
self.labels.each do |label|
|
257
|
-
new_data_frame.drop!(label) unless new_labels.include?(label)
|
258
|
-
end
|
259
|
-
new_data_frame
|
260
|
-
end
|
261
|
-
|
262
|
-
# A weird name. This creates a column for every category in a column
|
263
|
-
# and marks each row by its value
|
264
|
-
def j_binary_ize!(*columns)
|
265
|
-
# Allows to mix a hash with the columns.
|
266
|
-
options = columns.find_all {|e| e.is_a?(Hash)}.inject({}) {|h, e| h.merge!(e)}
|
267
|
-
columns.delete_if {|e| e.is_a?(Hash)}
|
268
|
-
|
269
|
-
# Generates new columns
|
270
|
-
columns.each do |col|
|
271
|
-
values = render_column(col.to_underscore_sym)
|
272
|
-
values.categories.each do |category|
|
273
|
-
full_name = (col.to_s + "_" + category.to_s).to_sym
|
274
|
-
if options[:allow_overlap]
|
275
|
-
category_map = values.inject([]) do |list, e|
|
276
|
-
list << values.all_categories(e)
|
277
|
-
end
|
278
|
-
self.append!(full_name, category_map.map{|e| e.include?(category)})
|
279
|
-
else
|
280
|
-
self.append!(full_name, values.category_map.map{|e| e == category})
|
281
|
-
end
|
282
|
-
end
|
283
|
-
end
|
284
|
-
end
|
285
|
-
|
286
|
-
# Adds a unique column to the table
|
287
|
-
def append!(column_name, value=nil)
|
288
|
-
raise ArgumentError, "Can't have duplicate column names" if self.labels.include?(column_name)
|
289
|
-
self.labels << column_name.to_underscore_sym
|
290
|
-
if value.is_a?(Array)
|
291
|
-
self.items.each_with_index do |item, i|
|
292
|
-
item << value[i]
|
293
|
-
end
|
294
|
-
else
|
295
|
-
self.items.each do |item|
|
296
|
-
item << value
|
297
|
-
end
|
298
|
-
end
|
299
|
-
# Because we are tainting the sub arrays, the TaintableArray doesn't know it's been changed.
|
300
|
-
self.items.taint
|
301
|
-
end
|
302
|
-
|
303
|
-
def filter_by_category(hash)
|
304
|
-
new_data_frame = self.dup
|
305
|
-
hash.each do |key, value|
|
306
|
-
key = key.to_underscore_sym
|
307
|
-
next unless self.labels.include?(key)
|
308
|
-
value = [value] unless value.is_a?(Array) or value.is_a?(Range)
|
309
|
-
new_data_frame.filter!(:hash) {|row| value.include?(row[key])}
|
310
|
-
end
|
311
|
-
new_data_frame
|
312
|
-
end
|
313
|
-
|
314
|
-
def filter_by_category!(hash)
|
315
|
-
hash.each do |key, value|
|
316
|
-
key = key.to_underscore_sym
|
317
|
-
next unless self.labels.include?(key)
|
318
|
-
value = [value] unless value.is_a?(Array) or value.is_a?(Range)
|
319
|
-
self.filter!(:hash) {|row| value.include?(row[key])}
|
320
|
-
end
|
321
|
-
end
|
322
|
-
|
323
|
-
end
|
22
|
+
require 'data_frame/parameter_capture'
|
23
|
+
require 'data_frame/data_frame'
|
24
|
+
require 'data_frame/model'
|
@@ -0,0 +1,301 @@
|
|
1
|
+
# This allows me to have named columns and optionally named rows in a
|
2
|
+
# data frame, to work calculations (usually on the columns), to
|
3
|
+
# transpose the matrix and store the transposed matrix until the object
|
4
|
+
# is tainted.
|
5
|
+
class DataFrame
|
6
|
+
|
7
|
+
class << self
|
8
|
+
|
9
|
+
# This is the neatest part of this neat gem.
|
10
|
+
# DataFrame.from_csv can be called in a lot of ways:
|
11
|
+
# DataFrame.from_csv(csv_contents)
|
12
|
+
# DataFrame.from_csv(filename)
|
13
|
+
# DataFrame.from_csv(url)
|
14
|
+
# If you need to define converters for FasterCSV, do it before calling
|
15
|
+
# this method:
|
16
|
+
# FasterCSV::Converters[:special] = lambda{|f| f == 'foo' ? 'bar' : 'foo'}
|
17
|
+
# DataFrame.from_csv('http://example.com/my_special_url.csv', :converters => :special)
|
18
|
+
# This returns bar where 'foo' was found and 'foo' everywhere else.
|
19
|
+
def from_csv(obj, opts={})
|
20
|
+
labels, table = infer_csv_contents(obj, opts)
|
21
|
+
return nil unless labels and table
|
22
|
+
df = new(*labels)
|
23
|
+
df.import(table)
|
24
|
+
df
|
25
|
+
end
|
26
|
+
|
27
|
+
protected
|
28
|
+
def infer_csv_contents(obj, opts={})
|
29
|
+
contents = File.read(obj) if File.exist?(obj)
|
30
|
+
begin
|
31
|
+
open(obj) {|f| contents = f.read} unless contents
|
32
|
+
rescue
|
33
|
+
nil
|
34
|
+
end
|
35
|
+
contents ||= obj if obj.is_a?(String)
|
36
|
+
return nil unless contents
|
37
|
+
table = FCSV.parse(contents, default_csv_opts.merge(opts))
|
38
|
+
labels = table.shift
|
39
|
+
while table.last.empty?
|
40
|
+
table.pop
|
41
|
+
end
|
42
|
+
[labels, table]
|
43
|
+
end
|
44
|
+
|
45
|
+
def default_csv_opts; {:converters => :all}; end
|
46
|
+
end
|
47
|
+
|
48
|
+
# Loads a batch of rows. Expects an array of arrays, else you don't
|
49
|
+
# know what you have.
|
50
|
+
def import(rows)
|
51
|
+
rows.each do |row|
|
52
|
+
self.add_item(row)
|
53
|
+
end
|
54
|
+
end
|
55
|
+
|
56
|
+
def inspect
|
57
|
+
"DataFrame rows: #{self.rows.size} labels: #{self.labels.inspect}"
|
58
|
+
end
|
59
|
+
|
60
|
+
# The labels of the data items
|
61
|
+
attr_reader :labels
|
62
|
+
alias :variables :labels
|
63
|
+
|
64
|
+
# The items stored in the frame
|
65
|
+
attr_reader :items
|
66
|
+
|
67
|
+
def initialize(*labels)
|
68
|
+
@labels = labels.map {|e| e.to_underscore_sym }
|
69
|
+
@items = TransposableArray.new
|
70
|
+
end
|
71
|
+
|
72
|
+
def add_item(item)
|
73
|
+
self.items << item
|
74
|
+
end
|
75
|
+
alias :add :add_item
|
76
|
+
|
77
|
+
def row_labels
|
78
|
+
@row_labels ||= []
|
79
|
+
end
|
80
|
+
|
81
|
+
def row_labels=(ary)
|
82
|
+
raise ArgumentError, "Row labels must be an array" unless ary.is_a?(Array)
|
83
|
+
@row_labels = ary
|
84
|
+
end
|
85
|
+
|
86
|
+
def render_column(sym)
|
87
|
+
i = @labels.index(sym)
|
88
|
+
return nil unless i
|
89
|
+
@items.transpose[i]
|
90
|
+
end
|
91
|
+
|
92
|
+
# The rows as an array of arrays, an alias for items.
|
93
|
+
alias :rows :items
|
94
|
+
|
95
|
+
# The columns as a Dictionary or Hash
|
96
|
+
# This is cached, call columns(true) to reset the cache.
|
97
|
+
def columns(reset=false)
|
98
|
+
@columns = nil if reset
|
99
|
+
return @columns if @columns
|
100
|
+
|
101
|
+
container = defined?(Dictionary) ? Dictionary.new : Hash.new
|
102
|
+
i = 0
|
103
|
+
|
104
|
+
@columns = @items.transpose.inject(container) do |cont, col|
|
105
|
+
cont[@labels[i]] = col
|
106
|
+
i += 1
|
107
|
+
cont
|
108
|
+
end
|
109
|
+
end
|
110
|
+
alias :to_hash :columns
|
111
|
+
alias :to_dictionary :columns
|
112
|
+
|
113
|
+
def render_row(sym)
|
114
|
+
i = self.row_labels.index(sym)
|
115
|
+
return nil unless i
|
116
|
+
@items[i]
|
117
|
+
end
|
118
|
+
|
119
|
+
def method_missing(sym, *args, &block)
|
120
|
+
if self.labels.include?(sym)
|
121
|
+
render_column(sym)
|
122
|
+
elsif self.row_labels.include?(sym)
|
123
|
+
render_row(sym)
|
124
|
+
elsif @items.respond_to?(sym)
|
125
|
+
@items.send(sym, *args, &block)
|
126
|
+
else
|
127
|
+
super
|
128
|
+
end
|
129
|
+
end
|
130
|
+
|
131
|
+
def drop!(*labels)
|
132
|
+
labels.each do |label|
|
133
|
+
drop_one!(label)
|
134
|
+
end
|
135
|
+
self
|
136
|
+
end
|
137
|
+
|
138
|
+
def drop_one!(label)
|
139
|
+
i = self.labels.index(label)
|
140
|
+
return nil unless i
|
141
|
+
self.items.each do |item|
|
142
|
+
item.delete_at(i)
|
143
|
+
end
|
144
|
+
self.labels.delete_at(i)
|
145
|
+
self
|
146
|
+
end
|
147
|
+
protected :drop_one!
|
148
|
+
|
149
|
+
def replace!(column, values=nil, &block)
|
150
|
+
column = validate_column(column)
|
151
|
+
if not values
|
152
|
+
values = self.send(column)
|
153
|
+
values.map! {|e| block.call(e)}
|
154
|
+
end
|
155
|
+
replace_column(column, values)
|
156
|
+
self
|
157
|
+
end
|
158
|
+
|
159
|
+
def replace_column(column, values)
|
160
|
+
column = validate_column(column)
|
161
|
+
index = self.labels.index(column)
|
162
|
+
list = []
|
163
|
+
self.items.each_with_index do |item, i|
|
164
|
+
consolidated = item
|
165
|
+
consolidated[index] = values[i]
|
166
|
+
list << consolidated
|
167
|
+
end
|
168
|
+
@items = list.dup
|
169
|
+
end
|
170
|
+
protected :replace_column
|
171
|
+
|
172
|
+
def validate_column(column)
|
173
|
+
column = column.to_sym
|
174
|
+
raise ArgumentError, "Must provide the name of an existing column. Provided #{column.inspect}, needed to provide one of #{self.labels.inspect}" unless self.labels.include?(column)
|
175
|
+
column
|
176
|
+
end
|
177
|
+
protected :validate_column
|
178
|
+
|
179
|
+
# Takes a block to evaluate on each row. The row can be converted into
|
180
|
+
# an OpenStruct or a Hash for easier filter methods. Note, don't try this
|
181
|
+
# with a hash or open struct unless you have facets available.
|
182
|
+
def filter!(as=Array, &block)
|
183
|
+
as = infer_class(as)
|
184
|
+
items = []
|
185
|
+
self.items.each do |row|
|
186
|
+
value = block.call(cast_row(row, as))
|
187
|
+
items << row if value
|
188
|
+
end
|
189
|
+
@items = items.dup
|
190
|
+
self
|
191
|
+
end
|
192
|
+
|
193
|
+
def filter(as=Array, &block)
|
194
|
+
new_data_frame = self.clone
|
195
|
+
new_data_frame.filter!(as, &block)
|
196
|
+
end
|
197
|
+
|
198
|
+
def infer_class(obj)
|
199
|
+
obj = obj.to_s.classify.constantize if obj.is_a?(Symbol)
|
200
|
+
obj = obj.classify.constantize if obj.is_a?(String)
|
201
|
+
obj
|
202
|
+
end
|
203
|
+
protected :infer_class
|
204
|
+
|
205
|
+
def cast_row(row, as)
|
206
|
+
if as == Hash
|
207
|
+
obj = {}
|
208
|
+
self.labels.each_with_index do |label, i|
|
209
|
+
obj[label] = row[i]
|
210
|
+
end
|
211
|
+
obj
|
212
|
+
elsif as == OpenStruct
|
213
|
+
obj = OpenStruct.new
|
214
|
+
self.labels.each_with_index do |label, i|
|
215
|
+
obj.table[label] = row[i]
|
216
|
+
end
|
217
|
+
obj
|
218
|
+
elsif as == Array
|
219
|
+
row
|
220
|
+
else
|
221
|
+
as.new(*row)
|
222
|
+
end
|
223
|
+
end
|
224
|
+
protected :cast_row
|
225
|
+
|
226
|
+
# Creates a new data frame, only with the specified columns.
|
227
|
+
def subset_from_columns(*cols)
|
228
|
+
new_labels = self.labels.inject([]) do |list, label|
|
229
|
+
list << label if cols.include?(label)
|
230
|
+
list
|
231
|
+
end
|
232
|
+
new_data_frame = DataFrame.new(*self.labels)
|
233
|
+
new_data_frame.import(self.items)
|
234
|
+
self.labels.each do |label|
|
235
|
+
new_data_frame.drop!(label) unless new_labels.include?(label)
|
236
|
+
end
|
237
|
+
new_data_frame
|
238
|
+
end
|
239
|
+
|
240
|
+
# A weird name. This creates a column for every category in a column
|
241
|
+
# and marks each row by its value
|
242
|
+
def j_binary_ize!(*columns)
|
243
|
+
# Allows to mix a hash with the columns.
|
244
|
+
options = columns.find_all {|e| e.is_a?(Hash)}.inject({}) {|h, e| h.merge!(e)}
|
245
|
+
columns.delete_if {|e| e.is_a?(Hash)}
|
246
|
+
|
247
|
+
# Generates new columns
|
248
|
+
columns.each do |col|
|
249
|
+
values = render_column(col.to_underscore_sym)
|
250
|
+
values.categories.each do |category|
|
251
|
+
full_name = (col.to_s + "_" + category.to_s).to_sym
|
252
|
+
if options[:allow_overlap]
|
253
|
+
category_map = values.inject([]) do |list, e|
|
254
|
+
list << values.all_categories(e)
|
255
|
+
end
|
256
|
+
self.append!(full_name, category_map.map{|e| e.include?(category)})
|
257
|
+
else
|
258
|
+
self.append!(full_name, values.category_map.map{|e| e == category})
|
259
|
+
end
|
260
|
+
end
|
261
|
+
end
|
262
|
+
end
|
263
|
+
|
264
|
+
# Adds a unique column to the table
|
265
|
+
def append!(column_name, value=nil)
|
266
|
+
raise ArgumentError, "Can't have duplicate column names" if self.labels.include?(column_name)
|
267
|
+
self.labels << column_name.to_underscore_sym
|
268
|
+
if value.is_a?(Array)
|
269
|
+
self.items.each_with_index do |item, i|
|
270
|
+
item << value[i]
|
271
|
+
end
|
272
|
+
else
|
273
|
+
self.items.each do |item|
|
274
|
+
item << value
|
275
|
+
end
|
276
|
+
end
|
277
|
+
# Because we are tainting the sub arrays, the TaintableArray doesn't know it's been changed.
|
278
|
+
self.items.taint
|
279
|
+
end
|
280
|
+
|
281
|
+
def filter_by_category(hash)
|
282
|
+
new_data_frame = self.dup
|
283
|
+
hash.each do |key, value|
|
284
|
+
key = key.to_underscore_sym
|
285
|
+
next unless self.labels.include?(key)
|
286
|
+
value = [value] unless value.is_a?(Array) or value.is_a?(Range)
|
287
|
+
new_data_frame.filter!(:hash) {|row| value.include?(row[key])}
|
288
|
+
end
|
289
|
+
new_data_frame
|
290
|
+
end
|
291
|
+
|
292
|
+
def filter_by_category!(hash)
|
293
|
+
hash.each do |key, value|
|
294
|
+
key = key.to_underscore_sym
|
295
|
+
next unless self.labels.include?(key)
|
296
|
+
value = [value] unless value.is_a?(Array) or value.is_a?(Range)
|
297
|
+
self.filter!(:hash) {|row| value.include?(row[key])}
|
298
|
+
end
|
299
|
+
end
|
300
|
+
|
301
|
+
end
|