/
class_decomposition.py
189 lines (165 loc) · 8.95 KB
/
class_decomposition.py
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
# ======================================== #
# Class that does the decomposition
# Input: files produced by class_datahandling.py
# Output: XRvar.nc, which contains the variance fractions explained by the three drivers
# ======================================== #
# =========================================================== #
# PREAMBULE
# Put in packages that we need
# =========================================================== #
from pathlib import Path
import numpy as np
from tqdm import tqdm
import pickle
import xarray as xr
import yaml
# =========================================================== #
# CLASS OBJECT
# =========================================================== #
class decomposition(object):
# =========================================================== #
# =========================================================== #
def __init__(self):
print("# ==================================== #")
print("# Initializing decomposition class #")
print("# ==================================== #")
self.current_dir = Path.cwd()
# Read in Input YAML file
with open(self.current_dir / 'input.yml') as file:
diction = yaml.load(file, Loader=yaml.FullLoader)
self.location_ipccdata = Path(diction['location_ipcc'])
self.save = diction['save']
self.sample_size_per_ms = int(diction['sample_size_per_ms'])
self.resampling = int(diction['resampling'])
self.removal_c8 = diction['removal_c8']
self.generate_composites = diction['generate_composites']
# Read in data produced by the code in data_handling.py
self.XRmeta = xr.open_dataset(self.current_dir / "Data" / "Handling_files" / ("XRmeta.nc"))
self.XRar6 = xr.open_dataset(self.current_dir / "Data" / "Handling_files" / ("XRdata.nc"))
self.years = list(self.XRar6.Time.data)
self.var_all = np.array(self.XRar6.Variable)
with open(self.current_dir / "Data" / "Handling_files" / 'XRsubs.pickle', 'rb') as handle:
self.XRsubs = pickle.load(handle)
# =========================================================== #
# =========================================================== #
def sampling_and_decomposing(self, printy='on'):
if printy=='on': print("- Generate samples and apply decomposition")
def generate_listoflists(self, xrsub):
xrmeta = self.XRmeta
modscens = np.array(xrsub.ModelScenario)
mods = np.array([i.split('|')[0] for i in modscens])
ccat = np.array(xrmeta.sel(ModelScenario=xrsub.ModelScenario).Category.data)
unimods = np.unique(mods)
uniccat = np.unique(ccat)
whs = []
for m_i, m in enumerate(unimods):
for c_i, c in enumerate(uniccat):
wh = np.where((mods == m) & (ccat == c))[0]
whs.append(wh)
return whs
def generate_samples(self, var):
xrsub = self.XRsubs[var]
values = np.array(xrsub.Value)
values_nn = np.array(xrsub.Value)
values = values - np.mean(values)
values = values / np.std(values)
modscens = np.array(xrsub.ModelScenario)
mods = np.array([i.split('|')[0] for i in modscens])
ccat = np.array(self.XRmeta.sel(ModelScenario=xrsub.ModelScenario).Category.data)
unimods = np.unique(mods)
uniccat = np.unique(ccat)
whs = generate_listoflists(self, xrsub)
ss = self.sample_size_per_ms
indices = np.zeros(shape=(7, self.resampling, len(self.years)))
for n_i in range(self.resampling):
sample1 = np.zeros(shape=(2, len(whs)*ss)).astype(str)
a = 0
for m_i, m in enumerate(unimods):
for c_i, c in enumerate(uniccat):
wh = whs[a]
sample1[0][a*ss:a*ss+ss] = [m]*ss
sample1[1][a*ss:a*ss+ss] = [c]*ss
a+=1
np.random.shuffle(sample1.T)
sample2 = np.zeros(shape=(2, len(whs)*ss)).astype(str)
a = 0
for m_i, m in enumerate(unimods):
for c_i, c in enumerate(uniccat):
wh = whs[a]
sample2[0][a*ss:a*ss+ss] = [m]*ss
sample2[1][a*ss:a*ss+ss] = [c]*ss
a+=1
np.random.shuffle(sample2.T)
M1 = np.zeros(shape=(len(sample1[0]), len(self.years)))
M1nn = np.zeros(shape=(len(sample1[0]), len(self.years)))
M2 = np.zeros(shape=(len(sample1[0]), len(self.years)))
Nm = np.zeros(shape=(len(sample1[0]), len(self.years)))
Nc = np.zeros(shape=(len(sample1[0]), len(self.years)))
Nmc = np.zeros(shape=(len(sample1[0]), len(self.years)))
cv = np.zeros(shape=(len(sample1[0]), len(self.years)))
for m in unimods:
for c in uniccat:
wh = np.where((mods == m) & (ccat == c))[0]
if len(wh) > 0:
wh1 = np.where((sample1[0] == m) & (sample1[1] == c))[0]
wh2 = np.where((sample2[0] == m) & (sample2[1] == c))[0]
choice = np.random.choice(wh, self.sample_size_per_ms, replace=True)
M1nn[wh1] = values_nn[choice]
M1[wh1] = values[choice]
M2[wh2] = values[np.random.choice(wh, self.sample_size_per_ms, replace=True)]
wh_m = np.where((sample1[0] == m) & (sample2[1] == c))[0]
wh_c = np.where((sample2[0] == m) & (sample1[1] == c))[0]
Nm[wh_m] = values[np.random.choice(wh, len(wh_m), replace=True)]
Nc[wh_c] = values[np.random.choice(wh, len(wh_c), replace=True)]
Nmc[wh1] = values[np.random.choice(wh, len(wh1), replace=True)]
vtot = np.var(M1nn, axis=0) / np.mean(M1nn)
vtot_norm = np.var(M1, axis=0)
cv = np.std(M1nn, axis=0) / np.mean(M1nn)
s_m = np.diag(1/(len(sample1[0])-1)*np.dot(M1.T, Nm) - 1/(len(sample1[0]))*np.dot(M1.T, M2))/np.var(M1, axis=0)
s_c = np.diag(1/(len(sample1[0])-1)*np.dot(M1.T, Nc) - 1/(len(sample1[0]))*np.dot(M1.T, M2))/np.var(M1, axis=0)
comb = np.diag(1/(len(sample1[0])-1)*np.dot(M1.T, Nmc) - 1/len(sample1[0])*np.dot(M1.T, M2))/np.var(M1, axis=0)
s_mc = comb - s_m - s_c
s_z = 1 - comb
indices[:, n_i, :] = [vtot, s_m, s_c, s_mc, s_z, vtot_norm, cv]
self.comb = comb
return np.mean(indices, axis=1)
self.variances = np.zeros(shape=(len(self.var_all), 7, len(self.years)))
if printy == 'on':
for v_i in tqdm(range(len(self.var_all))):
var = self.var_all[v_i]
vtot, s_m, s_c, s_mc, s_z, vtot_norm, cv = generate_samples(self, var)
self.variances[v_i][0] = vtot
self.variances[v_i][1] = s_m
self.variances[v_i][2] = s_c
self.variances[v_i][3] = s_z
self.variances[v_i][4] = s_mc
self.variances[v_i][5] = vtot_norm
self.variances[v_i][6] = cv
else:
for v_i in range(len(self.var_all)):
var = self.var_all[v_i]
vtot, s_m, s_c, s_mc, s_z, vtot_norm, cv = generate_samples(self, var)
self.variances[v_i][0] = vtot
self.variances[v_i][1] = s_m
self.variances[v_i][2] = s_c
self.variances[v_i][3] = s_z
self.variances[v_i][4] = s_mc
self.variances[v_i][5] = vtot_norm
self.variances[v_i][6] = cv
ds = xr.Dataset({"Var_total": (("Variable", "Time"), self.variances[:, 0]),
"S_m": (("Variable", "Time"), self.variances[:, 1]),
"S_c": (("Variable", "Time"), self.variances[:, 2]),
"S_z": (("Variable", "Time"), self.variances[:, 3]),
"S_mc": (("Variable", "Time"), self.variances[:, 4]),
"Var_total_norm": (("Variable", "Time"), self.variances[:, 5]),
"CoefVar": (("Variable", "Time"), self.variances[:, 6])},
coords={
"Variable": self.var_all,
"Time": self.years})
self.XRvar = ds
# =========================================================== #
# =========================================================== #
def savings(self):
print("- Save stuff")
if self.save == 'yes':
self.XRvar.to_netcdf(self.current_dir / "Data" / "Output_files" / "Variances.nc")