davidrichards-data_frame 0.0.18 → 0.0.19
Sign up to get free protection for your applications and to get access to all the features.
- data/README.rdoc +16 -0
- data/VERSION.yml +1 -1
- data/bin/plain_frame +22 -0
- data/lib/data_frame.rb +2 -1
- data/lib/data_frame/arff.rb +43 -36
- data/lib/data_frame/core/column_management.rb +102 -0
- data/lib/data_frame/core/filter.rb +48 -0
- data/lib/data_frame/core/import.rb +112 -0
- data/lib/data_frame/core/pre_process.rb +61 -0
- data/lib/data_frame/core/saving.rb +29 -0
- data/lib/data_frame/core/training.rb +36 -0
- data/lib/data_frame/data_frame.rb +37 -241
- data/lib/data_frame/id3.rb +28 -0
- data/lib/data_frame/kmeans.rb +10 -0
- data/lib/data_frame/labels_from_uci.rb +48 -0
- data/lib/data_frame/mlp.rb +18 -0
- data/lib/data_frame/sbn.rb +18 -0
- data/lib/data_frame/transposable_array.rb +1 -1
- data/lib/ext/array.rb +11 -0
- data/spec/data_frame/arff_spec.rb +1 -0
- data/spec/data_frame/core/column_management_spec.rb +97 -0
- data/spec/data_frame/core/filter_spec.rb +88 -0
- data/spec/data_frame/core/import_spec.rb +41 -0
- data/spec/data_frame/core/pre_process_spec.rb +71 -0
- data/spec/data_frame/core/saving_spec.rb +61 -0
- data/spec/data_frame/core/training_spec.rb +51 -0
- data/spec/data_frame/data_frame_spec.rb +10 -226
- data/spec/data_frame/id3_spec.rb +22 -0
- data/spec/ext/array_spec.rb +13 -0
- data/spec/fixtures/discrete_testing.csv +4 -0
- data/spec/fixtures/discrete_training.csv +21 -0
- metadata +33 -6
data/README.rdoc
CHANGED
@@ -91,6 +91,22 @@ Data Frame can now create sub-models:
|
|
91
91
|
>> df.models
|
92
92
|
=> #<OpenStruct weekend=DataFrame rows: 179 labels: [:x, :y, :month, :day, :ffmc, :dmc, :dc, :isi, :temp, :rh, :wind, :rain, :area]>
|
93
93
|
|
94
|
+
== Utilities
|
95
|
+
|
96
|
+
I use data frame for a lot of things, and I've added some utilities for this gem in case you would like to as well. For instance, here is how I take the data in a data frame and load it into a neural network:
|
97
|
+
|
98
|
+
# Show mlp. Will probably need to add a row classifier for training and test data. Also, will probably want to
|
99
|
+
|
100
|
+
== CLI
|
101
|
+
|
102
|
+
There are some really interesting things that have good command-line shortcuts:
|
103
|
+
|
104
|
+
* Make
|
105
|
+
* A
|
106
|
+
* List
|
107
|
+
|
108
|
+
# Now add some demos
|
109
|
+
|
94
110
|
==Installation
|
95
111
|
|
96
112
|
sudo gem install davidrichards-data_frame
|
data/VERSION.yml
CHANGED
data/bin/plain_frame
ADDED
@@ -0,0 +1,22 @@
|
|
1
|
+
#!/usr/bin/env ruby -wKU
|
2
|
+
require 'yaml'
|
3
|
+
|
4
|
+
version_hash = YAML.load_file(File.join(File.dirname(__FILE__), %w(.. VERSION.yml)))
|
5
|
+
version = [version_hash[:major].to_s, version_hash[:minor].to_s, version_hash[:patch].to_s].join(".")
|
6
|
+
df_file = File.join(File.dirname(__FILE__), %w(.. lib data_frame))
|
7
|
+
|
8
|
+
irb = RUBY_PLATFORM =~ /(:?mswin|mingw)/ ? 'irb.bat' : 'irb'
|
9
|
+
|
10
|
+
require 'optparse'
|
11
|
+
options = { :irb => irb, :without_stored_procedures => false }
|
12
|
+
OptionParser.new do |opt|
|
13
|
+
opt.banner = "Usage: console [environment] [options]"
|
14
|
+
opt.on("--irb=[#{irb}]", 'Invoke a different irb.') { |v| options[:irb] = v }
|
15
|
+
opt.parse!(ARGV)
|
16
|
+
end
|
17
|
+
|
18
|
+
libs = " -r irb/completion -r #{df_file}"
|
19
|
+
|
20
|
+
puts "Loading Data Frame version: #{version}"
|
21
|
+
|
22
|
+
exec "#{options[:irb]} #{libs} --simple-prompt"
|
data/lib/data_frame.rb
CHANGED
@@ -20,6 +20,7 @@ $:.unshift(File.dirname(__FILE__))
|
|
20
20
|
require 'data_frame/callback_array'
|
21
21
|
require 'data_frame/transposable_array'
|
22
22
|
require 'data_frame/parameter_capture'
|
23
|
-
require 'data_frame/arff'
|
24
23
|
require 'data_frame/data_frame'
|
25
24
|
require 'data_frame/model'
|
25
|
+
|
26
|
+
Dir.glob("#{File.dirname(__FILE__)}/data_frame/core/*.rb").each { |file| require file }
|
data/lib/data_frame/arff.rb
CHANGED
@@ -1,45 +1,52 @@
|
|
1
|
-
|
2
|
-
|
3
|
-
|
4
|
-
# Used in arff, but generally useful.
|
5
|
-
def to_csv(include_header=true)
|
6
|
-
value = include_header ? self.labels.map{|e| e.to_s}.join(',') + "\n" : ''
|
7
|
-
self.items.inject(value) do |list, e|
|
8
|
-
list << e.map {|cell| cell.to_s}.join(',') + "\n"
|
9
|
-
end
|
10
|
-
end
|
1
|
+
module DF #:nodoc:
|
2
|
+
# Turns a data frame into ARFF-formatted content.
|
3
|
+
module ARFF
|
11
4
|
|
12
|
-
|
13
|
-
|
14
|
-
|
15
|
-
|
16
|
-
|
17
|
-
def arff_attributes
|
18
|
-
container = defined?(Dictionary) ? Dictionary.new : Hash.new
|
19
|
-
|
20
|
-
self.labels.inject(container) do |list, e|
|
21
|
-
list[e] = self.render_column(e).categories
|
22
|
-
end
|
23
|
-
end
|
24
|
-
|
25
|
-
def arff_formatted_attributes
|
26
|
-
self.labels.inject('') do |str, e|
|
27
|
-
val = "{" + self.render_column(e).categories.map{|x| x.to_s}.join(',') + "}"
|
28
|
-
str << "@attribute #{e} #{val}\n"
|
5
|
+
# Used in arff, but generally useful.
|
6
|
+
def to_csv(include_header=true)
|
7
|
+
value = include_header ? self.labels.map{|e| e.to_s}.join(',') + "\n" : ''
|
8
|
+
self.items.inject(value) do |list, e|
|
9
|
+
list << e.map {|cell| cell.to_s}.join(',') + "\n"
|
29
10
|
end
|
30
11
|
end
|
31
|
-
|
32
|
-
def
|
33
|
-
|
12
|
+
|
13
|
+
def to_arff
|
14
|
+
arff_header + to_csv(false)
|
34
15
|
end
|
35
|
-
|
36
|
-
|
37
|
-
|
16
|
+
|
17
|
+
protected
|
18
|
+
def arff_attributes
|
19
|
+
container = defined?(Dictionary) ? Dictionary.new : Hash.new
|
20
|
+
|
21
|
+
self.labels.inject(container) do |list, e|
|
22
|
+
list[e] = self.render_column(e).categories
|
23
|
+
end
|
24
|
+
end
|
25
|
+
|
26
|
+
def arff_formatted_attributes
|
27
|
+
self.labels.inject('') do |str, e|
|
28
|
+
val = "{" + self.render_column(e).categories.map{|x| x.to_s}.join(',') + "}"
|
29
|
+
str << "@attribute #{e} #{val}\n"
|
30
|
+
end
|
31
|
+
end
|
32
|
+
|
33
|
+
def arff_relation
|
34
|
+
self.name ? self.name.to_underscore_sym.to_s : 'unamed_relation'
|
35
|
+
end
|
36
|
+
|
37
|
+
def arff_header
|
38
|
+
%[@relation #{arff_relation}
|
38
39
|
|
39
40
|
#{arff_formatted_attributes}
|
40
41
|
@data
|
41
42
|
]
|
42
|
-
|
43
|
-
|
44
|
-
|
43
|
+
end
|
44
|
+
|
45
|
+
alias :arff_items :to_csv
|
46
|
+
end
|
47
|
+
|
48
|
+
end
|
49
|
+
|
50
|
+
class DataFrame
|
51
|
+
include DF::ARFF
|
45
52
|
end
|
@@ -0,0 +1,102 @@
|
|
1
|
+
module DF #:nodoc:
|
2
|
+
module ColumnManagement #:nodoc:
|
3
|
+
|
4
|
+
def move_to_last!(orig_name)
|
5
|
+
raise ArgumentError, "Column not found" unless self.labels.include?(orig_name)
|
6
|
+
new_name = (orig_name.to_s + "_a_unique_name").to_sym
|
7
|
+
self.append!(new_name, self.render_column(orig_name))
|
8
|
+
self.drop!(orig_name)
|
9
|
+
self.rename!(orig_name, new_name)
|
10
|
+
end
|
11
|
+
|
12
|
+
# In the order of alias: new_name, orig_name
|
13
|
+
def rename!(new_name, orig_name)
|
14
|
+
new_name = new_name.to_underscore_sym
|
15
|
+
orig_name = orig_name.to_underscore_sym
|
16
|
+
raise ArgumentError, "Column not found" unless self.labels.include?(orig_name)
|
17
|
+
raise ArgumentError, "Cannot name #{orig_name} to #{new_name}, that column already exists." if self.labels.include?(new_name)
|
18
|
+
i = self.labels.index(orig_name)
|
19
|
+
self.labels[i] = new_name
|
20
|
+
end
|
21
|
+
|
22
|
+
# Adds a unique column to the table
|
23
|
+
def append!(column_name, value=nil)
|
24
|
+
raise ArgumentError, "Can't have duplicate column names" if self.labels.include?(column_name)
|
25
|
+
self.labels << column_name.to_underscore_sym
|
26
|
+
if value.is_a?(Array)
|
27
|
+
self.items.each_with_index do |item, i|
|
28
|
+
item << value[i]
|
29
|
+
end
|
30
|
+
else
|
31
|
+
self.items.each do |item|
|
32
|
+
item << value
|
33
|
+
end
|
34
|
+
end
|
35
|
+
# Because we are tainting the sub arrays, the TaintableArray doesn't know it's been changed.
|
36
|
+
self.items.taint
|
37
|
+
end
|
38
|
+
|
39
|
+
def replace!(column, values=nil, &block)
|
40
|
+
column = validate_column(column)
|
41
|
+
if not values
|
42
|
+
values = self.send(column)
|
43
|
+
values.map! {|e| block.call(e)}
|
44
|
+
end
|
45
|
+
replace_column!(column, values)
|
46
|
+
self
|
47
|
+
end
|
48
|
+
|
49
|
+
# Replace a single column with an array of values.
|
50
|
+
# It is helpful to have the values the same size as the rest of the data
|
51
|
+
# frame.
|
52
|
+
def replace_column!(column, values)
|
53
|
+
column = validate_column(column)
|
54
|
+
index = self.labels.index(column)
|
55
|
+
list = []
|
56
|
+
self.items.each_with_index do |item, i|
|
57
|
+
consolidated = item
|
58
|
+
consolidated[index] = values[i]
|
59
|
+
list << consolidated
|
60
|
+
end
|
61
|
+
@items = list.dup
|
62
|
+
end
|
63
|
+
|
64
|
+
# Drop one or more columns
|
65
|
+
def drop!(*labels)
|
66
|
+
labels.each do |label|
|
67
|
+
drop_one!(label)
|
68
|
+
end
|
69
|
+
self
|
70
|
+
end
|
71
|
+
|
72
|
+
# Drop a single column
|
73
|
+
def drop_one!(label)
|
74
|
+
i = self.labels.index(label)
|
75
|
+
return nil unless i
|
76
|
+
self.items.each do |item|
|
77
|
+
item.delete_at(i)
|
78
|
+
end
|
79
|
+
self.labels.delete_at(i)
|
80
|
+
self
|
81
|
+
end
|
82
|
+
|
83
|
+
# Creates a new data frame, only with the specified columns.
|
84
|
+
def subset_from_columns(*cols)
|
85
|
+
new_labels = self.labels.inject([]) do |list, label|
|
86
|
+
list << label if cols.include?(label)
|
87
|
+
list
|
88
|
+
end
|
89
|
+
new_data_frame = DataFrame.new(*self.labels)
|
90
|
+
new_data_frame.import(self.items)
|
91
|
+
self.labels.each do |label|
|
92
|
+
new_data_frame.drop!(label) unless new_labels.include?(label)
|
93
|
+
end
|
94
|
+
new_data_frame
|
95
|
+
end
|
96
|
+
|
97
|
+
end
|
98
|
+
end
|
99
|
+
|
100
|
+
class DataFrame
|
101
|
+
include DF::ColumnManagement
|
102
|
+
end
|
@@ -0,0 +1,48 @@
|
|
1
|
+
module DF #:nodoc:
|
2
|
+
module Filter #:nodoc:
|
3
|
+
|
4
|
+
# Takes a block to evaluate on each row. The row can be converted into
|
5
|
+
# an OpenStruct or a Hash for easier filter methods. Note, don't try this
|
6
|
+
# with a hash or open struct unless you have facets available.
|
7
|
+
def filter!(as=Array, &block)
|
8
|
+
as = infer_class(as)
|
9
|
+
items = []
|
10
|
+
self.items.each do |row|
|
11
|
+
value = block.call(cast_row(row, as))
|
12
|
+
items << row if value
|
13
|
+
end
|
14
|
+
@items = items.dup
|
15
|
+
self
|
16
|
+
end
|
17
|
+
|
18
|
+
def filter(as=Array, &block)
|
19
|
+
new_data_frame = self.clone
|
20
|
+
new_data_frame.filter!(as, &block)
|
21
|
+
end
|
22
|
+
|
23
|
+
def filter_by_category(hash)
|
24
|
+
new_data_frame = self.dup
|
25
|
+
hash.each do |key, value|
|
26
|
+
key = key.to_underscore_sym
|
27
|
+
next unless self.labels.include?(key)
|
28
|
+
value = [value] unless value.is_a?(Array) or value.is_a?(Range)
|
29
|
+
new_data_frame.filter!(:hash) {|row| value.include?(row[key])}
|
30
|
+
end
|
31
|
+
new_data_frame
|
32
|
+
end
|
33
|
+
|
34
|
+
def filter_by_category!(hash)
|
35
|
+
hash.each do |key, value|
|
36
|
+
key = key.to_underscore_sym
|
37
|
+
next unless self.labels.include?(key)
|
38
|
+
value = [value] unless value.is_a?(Array) or value.is_a?(Range)
|
39
|
+
self.filter!(:hash) {|row| value.include?(row[key])}
|
40
|
+
end
|
41
|
+
end
|
42
|
+
|
43
|
+
end
|
44
|
+
end
|
45
|
+
|
46
|
+
class DataFrame
|
47
|
+
include DF::Filter
|
48
|
+
end
|
@@ -0,0 +1,112 @@
|
|
1
|
+
module DF #:nodoc:
|
2
|
+
module Import #:nodoc:
|
3
|
+
|
4
|
+
module InferCSV #:nodoc:
|
5
|
+
|
6
|
+
protected
|
7
|
+
def default_csv_opts; {:converters => :all}; end
|
8
|
+
|
9
|
+
def infer_csv_contents(obj, opts={})
|
10
|
+
contents = File.read(obj) if File.exist?(obj)
|
11
|
+
begin
|
12
|
+
open(obj) {|f| contents = f.read} unless contents
|
13
|
+
rescue
|
14
|
+
nil
|
15
|
+
end
|
16
|
+
contents ||= obj if obj.is_a?(String)
|
17
|
+
return nil unless contents
|
18
|
+
table = FCSV.parse(contents, default_csv_opts.merge(opts))
|
19
|
+
labels = opts.fetch(:headers, true) ? table.shift : []
|
20
|
+
while table.last.empty?
|
21
|
+
table.pop
|
22
|
+
end
|
23
|
+
[labels, table]
|
24
|
+
end
|
25
|
+
|
26
|
+
end # InferCSV
|
27
|
+
|
28
|
+
module ClassMethods #:nodoc:
|
29
|
+
|
30
|
+
include InferCSV
|
31
|
+
|
32
|
+
# This is the neatest part of this neat gem.
|
33
|
+
# DataFrame.from_csv can be called in a lot of ways:
|
34
|
+
# DataFrame.from_csv(csv_contents)
|
35
|
+
# DataFrame.from_csv(filename)
|
36
|
+
# DataFrame.from_csv(url)
|
37
|
+
# If you need to define converters for FasterCSV, do it before calling
|
38
|
+
# this method:
|
39
|
+
# FasterCSV::Converters[:special] = lambda{|f| f == 'foo' ? 'bar' : 'foo'}
|
40
|
+
# DataFrame.from_csv('http://example.com/my_special_url.csv', :converters => :special)
|
41
|
+
# This returns bar where 'foo' was found and 'foo' everywhere else.
|
42
|
+
def from_csv(obj, opts={})
|
43
|
+
labels, table = infer_csv_contents(obj, opts)
|
44
|
+
name = infer_name_from_contents(obj, opts)
|
45
|
+
return nil unless labels and table
|
46
|
+
df = new(*labels)
|
47
|
+
df.import(table)
|
48
|
+
df.name = name
|
49
|
+
df
|
50
|
+
end
|
51
|
+
|
52
|
+
protected
|
53
|
+
|
54
|
+
# Only works for names sources, urls and files
|
55
|
+
def infer_name_from_contents(obj, opts={})
|
56
|
+
begin
|
57
|
+
File.split(obj).last.split('.')[0..-2].join('.').titleize
|
58
|
+
rescue
|
59
|
+
nil
|
60
|
+
end
|
61
|
+
end
|
62
|
+
|
63
|
+
end # Class Methods
|
64
|
+
|
65
|
+
module InstanceMethods #:nodoc:
|
66
|
+
|
67
|
+
include InferCSV
|
68
|
+
|
69
|
+
def add_item(item)
|
70
|
+
self.items << item
|
71
|
+
end
|
72
|
+
alias :add :add_item
|
73
|
+
|
74
|
+
# Loads a batch of rows. Expects an array of arrays, else you don't
|
75
|
+
# know what you have.
|
76
|
+
def import(rows)
|
77
|
+
case rows
|
78
|
+
when Array
|
79
|
+
import_array(rows)
|
80
|
+
when String
|
81
|
+
labels, table = infer_csv_contents(rows, :headers => false)
|
82
|
+
import(table)
|
83
|
+
else
|
84
|
+
raise ArgumentError, "Don't know how to import data from #{rows.class}"
|
85
|
+
end
|
86
|
+
true
|
87
|
+
end
|
88
|
+
|
89
|
+
protected
|
90
|
+
# Imports a table as an array of arrays.
|
91
|
+
# If the array is one-dimensional and there is more than one label, it
|
92
|
+
# imports only one row.
|
93
|
+
def import_array(rows)
|
94
|
+
raise ArgumentError, "Can only work with arrays" unless rows.is_a?(Array)
|
95
|
+
if self.labels.size > 1 and rows.dimensions == 1
|
96
|
+
self.add_item(rows)
|
97
|
+
else
|
98
|
+
rows.each do |row|
|
99
|
+
self.add_item(row)
|
100
|
+
end
|
101
|
+
end
|
102
|
+
end
|
103
|
+
|
104
|
+
end # Instance Methods
|
105
|
+
|
106
|
+
end
|
107
|
+
end
|
108
|
+
|
109
|
+
class DataFrame
|
110
|
+
include DF::Import::InstanceMethods
|
111
|
+
extend DF::Import::ClassMethods
|
112
|
+
end
|
@@ -0,0 +1,61 @@
|
|
1
|
+
module DF #:nodoc:
|
2
|
+
module PreProcess #:nodoc:
|
3
|
+
# A weird name. This creates a column for every category in a column
|
4
|
+
# and marks each row by its value
|
5
|
+
def j_binary_ize!(*columns)
|
6
|
+
# Allows to mix a hash with the columns.
|
7
|
+
options = columns.find_all {|e| e.is_a?(Hash)}.inject({}) {|h, e| h.merge!(e)}
|
8
|
+
columns.delete_if {|e| e.is_a?(Hash)}
|
9
|
+
|
10
|
+
# Generates new columns
|
11
|
+
columns.each do |col|
|
12
|
+
values = render_column(col.to_underscore_sym)
|
13
|
+
values.categories.each do |category|
|
14
|
+
full_name = (col.to_s + "_" + category.to_s).to_sym
|
15
|
+
if options[:allow_overlap]
|
16
|
+
category_map = values.inject([]) do |list, e|
|
17
|
+
list << values.all_categories(e)
|
18
|
+
end
|
19
|
+
self.append!(full_name, category_map.map{|e| e.include?(category)})
|
20
|
+
else
|
21
|
+
self.append!(full_name, values.category_map.map{|e| e == category})
|
22
|
+
end
|
23
|
+
end
|
24
|
+
end
|
25
|
+
end
|
26
|
+
|
27
|
+
# Adds a column, numerical_column_name that shows the same data as a
|
28
|
+
# nominal value, but as a number.
|
29
|
+
def numericize!(*columns)
|
30
|
+
columns.each do |col|
|
31
|
+
values = render_column(col.to_underscore_sym)
|
32
|
+
categories = values.categories
|
33
|
+
value_categories = values.map {|v| values.category(v)}
|
34
|
+
|
35
|
+
i = 0
|
36
|
+
category_map = value_categories.uniq.inject({}) do |h, c|
|
37
|
+
h[c] = i
|
38
|
+
i += 1
|
39
|
+
h
|
40
|
+
end
|
41
|
+
|
42
|
+
blank = Array.new(category_map.size, 0)
|
43
|
+
reverse_category_map = category_map.inject({}) {|h, e| h[e.last] = e.first; h}
|
44
|
+
|
45
|
+
new_values = values.inject([]) do |list, val|
|
46
|
+
a = blank.dup
|
47
|
+
a[category_map[values.category(val)]] = 1
|
48
|
+
list << a
|
49
|
+
end
|
50
|
+
|
51
|
+
new_name = "numerical #{col.to_s}".to_underscore_sym
|
52
|
+
self.append!(new_name, new_values)
|
53
|
+
end
|
54
|
+
end
|
55
|
+
|
56
|
+
end
|
57
|
+
end
|
58
|
+
|
59
|
+
class DataFrame
|
60
|
+
include DF::PreProcess
|
61
|
+
end
|