-
Notifications
You must be signed in to change notification settings - Fork 20
/
normalizer.jl
197 lines (164 loc) · 5.17 KB
/
normalizer.jl
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
module Normalizers
import StatsBase
using StatsBase: zscore, ZScoreTransform,UnitRangeTransform
using Dates
using DataFrames: DataFrame
using Statistics
import MultivariateStats
using Random
using ..AbsTypes
using ..Utils
import ..AbsTypes: fit, fit!, transform, transform!
export fit, fit!,transform, transform!
export Normalizer
const MV=MultivariateStats
const gmethods = [:zscore,:unitrange,:sqrt,:log,:pca,:ppca,:fa]
"""
Normalizer(Dict(
:method => :zscore
))
Transforms continuous features into normalized form such as zscore, unitrange, square-root, log, pca, ppca
with parameter:
- `:method` => `:zscore` or `:unitrange` or `:sqrt` or `:log` or `pca` or `ppca` or `fa`
- `:zscore` => standard z-score with centering and scaling
- `:unitrange` => unit range normalization with centering and scaling
- `:sqrt` => square-root transform
- `:pca` => principal component analysis transform
- `:ppca` => probabilistic pca
- `:fa` => factor analysis
- `:log` => log transform
Example:
function generatedf()
Random.seed!(123)
gdate = DateTime(2014,1,1):Dates.Minute(15):DateTime(2016,1,1)
gval1 = rand(length(gdate))
gval2 = rand(length(gdate))
gval3 = rand(length(gdate))
X = DataFrame(Date=gdate,Value1=gval1,Value2=gval2,Value3=gval3)
X
end
X = generatedf()
norm = Normalizer(Dict(:method => :zscore))
fit!(norm,X)
res=transform!(norm,X)
Implements: `fit!`, `transform!`
"""
mutable struct Normalizer <: Transformer
name::String
model::Dict{Symbol,Any}
function Normalizer(args=Dict())
default_args = Dict(
:name => "nrmlzr",
:method => :zscore
)
cargs=nested_dict_merge(default_args,args)
cargs[:method] in gmethods || throw(ArgmentError("invalid method"))
cargs[:name] = cargs[:name]*"_"*randstring(3)
new(cargs[:name],cargs)
end
end
function Normalizer(st::Symbol)
Normalizer(Dict(:method=>st))
end
"""
fit!(st::Statifier, features::T, labels::Vector=[])
Validate argument features other than dates are continuous.
"""
function fit!(norm::Normalizer, features::DataFrame, labels::Vector=[])::Nothing
# check features are in correct format and no categorical values
(infer_eltype(features[:,1]) <: DateTime && infer_eltype(Matrix(features[:,2:end])) <: Real) ||
(infer_eltype(Matrix(features)) <: Real) ||
throw(ArgmentError("Normalizer.fit!: make sure features are purely float values or float values with Date on first column"))
return nothing
end
function fit(norm::Normalizer, features::DataFrame, labels::Vector=[])::Normalizer
fit!(norm,features,labels)
return deepcopy(norm)
end
"""
transform!(norm::Normalizer, features::T) where {T<:Union{Vector,Matrix,DataFrame}}
Compute statistics.
"""
function transform!(norm::Normalizer, pfeatures::DataFrame)::DataFrame
pfeatures != DataFrame() || return DataFrame()
res = Array{Float64,2}(undef,0,0)
if (infer_eltype(pfeatures[:,1]) <: DateTime && infer_eltype(Matrix(pfeatures[:,2:end])) <: Real)
res = processnumeric(norm,Matrix{Float64}(pfeatures[:,2:end])) |> x->DataFrame(x,:auto)
elseif infer_eltype(Matrix(pfeatures)) <: Real
features = pfeatures |> Array{Float64}
res = processnumeric(norm,features) |> x->DataFrame(x,:auto)
else
error("Normalizer.transform!: make sure features are purely float values or float values with Date on first column")
end
res
end
function transform(norm::Normalizer, pfeatures::DataFrame)::DataFrame
return transform!(norm,pfeatures)
end
function processnumeric(norm::Normalizer,features::Matrix)
if norm.model[:method] == :zscore
ztransform(features)
elseif norm.model[:method] == :unitrange
unitrtransform(features)
elseif norm.model[:method] == :pca
pca(features)
elseif norm.model[:method] == :ppca
ppca(features)
elseif norm.model[:method] == :ica
ica(features)
elseif norm.model[:method] == :fa
fa(features)
elseif norm.model[:method] == :sqrt
sqrtf(features)
elseif norm.model[:method] == :log
logf(features)
else
error("arg's :method is mapped to unknown keyword")
end
end
# apply sqrt transform
function sqrtf(X)
sqrt.(X)
end
# apply log transform
function logf(X)
log.(X)
end
# apply z-score transform
function ztransform(X)
xp = X' |> collect |> Matrix{Float64}
StatsBase.fit(ZScoreTransform, xp,dims=2; center=true, scale=true) |> dt -> StatsBase.transform(dt,xp)' |> collect
end
# unit-range
function unitrtransform(X)
xp = X' |> collect |> Matrix{Float64}
StatsBase.fit(UnitRangeTransform,xp,dims=2) |> dt -> StatsBase.transform(dt,xp)' |> collect
end
# pca
function pca(X)
xp = X' |> collect |> Matrix{Float64}
m = MV.fit(MV.PCA,xp)
MV.predict(m,xp)' |> collect
end
function ica(X,kk::Int=0)
k = kk
if k == 0
k = size(X)[2]
end
xp = X' |> collect |> Matrix{Float64}
m = MV.fit(MV.ICA,xp,k)
MV.predict(m,xp)' |> collect
end
# ppca
function ppca(X)
xp = X' |> collect |> Matrix{Float64}
m = MV.fit(MV.PPCA,xp)
MV.predict(m,xp)' |> collect
end
# fa
function fa(X)
xp = X' |> collect |> Matrix{Float64}
m = MV.fit(MV.FactorAnalysis,xp)
MV.predict(m,xp)' |> collect
end
end