-
Notifications
You must be signed in to change notification settings - Fork 0
/
cle_lstm_lagged.py
301 lines (262 loc) · 11.1 KB
/
cle_lstm_lagged.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
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
# %matplotlib inline
import scipy.cluster.hierarchy as shc
import seaborn as sns
import math
from sklearn.preprocessing import StandardScaler
from sklearn.preprocessing import MinMaxScaler
from sklearn.preprocessing import normalize
from numpy import unique
from numpy import where
from statsmodels.graphics.tsaplots import plot_acf
from statsmodels.graphics.tsaplots import plot_pacf
import matplotlib as mp
from sklearn.preprocessing import StandardScaler
from sklearn.decomposition import PCA
import math
from sklearn.metrics import mean_squared_error
from sklearn.model_selection import KFold
from tqdm import tqdm
from statistics import mean
from sklearn.metrics import mean_absolute_error
from sklearn.metrics import r2_score
df1 = pd.read_csv('./climatic_indices/AO.txt', delimiter=',', header=None)
df2 = pd.read_csv('./climatic_indices/EPO.txt', delimiter=',', header=None)
df3 = pd.read_csv('./climatic_indices/NAO.txt', delimiter=',', header=None)
df4 = pd.read_csv('./climatic_indices/NINO12.txt', delim_whitespace=True, header=None)
df5 = pd.read_csv('./climatic_indices/NINO3.txt', delim_whitespace=True, header=None)
df6 = pd.read_csv('./climatic_indices/NINO34.txt', delim_whitespace=True, header=None)
df7 = pd.read_csv('./climatic_indices/NINO4.txt', delim_whitespace=True, header=None)
df8 = pd.read_csv('./climatic_indices/PNA.txt', delimiter=',', header=None)
df9 = pd.read_csv('./climatic_indices/SOI.txt', delim_whitespace=True)
df10 = pd.read_csv('./climatic_indices/WPO.txt', delimiter=',', header=None)
cle_inf = pd.read_excel('./data/cle/CLE_INFLOW.xlsx')
cle_eva = pd.read_excel('./data/cle/CLE_EVAPORATION.xlsx')
cle_pre = pd.read_excel('./data/cle/CLE_PRECIPITATION.xlsx')
def convert_single_to_datetime(df):
df.iloc[:, 0] = df.iloc[:, 0].astype(str)
df['date'] = pd.to_datetime(df.iloc[:, 0], format='%Y%m%d', errors='coerce')
df.drop(df.columns[0], axis=1, inplace=True)
return df
def convert_double_to_datetime(df):
df['date_str'] = df.iloc[:, 0].astype(str) + df.iloc[:, 1].astype(str).str.zfill(3)
df['date'] = pd.to_datetime(df['date_str'], format='%Y%j', errors='coerce')
df.drop(df.columns[[0, 1]], axis=1, inplace=True)
df.drop('date_str', axis=1, inplace=True)
return df
def convert_triple_to_datetime(df):
df['date_str'] = df.iloc[:, 0].astype(str) + df.iloc[:, 1].astype(str).str.zfill(2) + df.iloc[:, 2].astype(str).str.zfill(2)
df['date'] = pd.to_datetime(df['date_str'], format='%Y%m%d', errors='coerce')
df.drop(df.columns[[0, 1, 2]], axis=1, inplace=True)
df.drop('date_str', axis=1, inplace=True)
return df
def convert_column_to_datetime(df, column_name):
df[column_name] = df[column_name].astype(str)
df['date'] = pd.to_datetime(df[column_name], format='%Y%m%d', errors='coerce')
df.drop(columns=[column_name], inplace=True)
return df
df1 = convert_triple_to_datetime(df1)
df2 = convert_triple_to_datetime(df2)
df3 = convert_triple_to_datetime(df3)
df4 = convert_single_to_datetime(df4)
df5 = convert_single_to_datetime(df5)
df6 = convert_single_to_datetime(df6)
df7 = convert_single_to_datetime(df7)
df8 = convert_triple_to_datetime(df8)
df9 = convert_double_to_datetime(df9)
df10 = convert_triple_to_datetime(df10)
cle_inf = convert_column_to_datetime(cle_inf, 'OBS DATE')
cle_eva = convert_column_to_datetime(cle_eva, 'OBS DATE')
cle_pre = convert_column_to_datetime(cle_pre, 'OBS DATE')
def find_min_max_dates(dataframes):
min_date = dataframes[0]['date'].min()
max_date = dataframes[0]['date'].max()
for df in dataframes:
if 'date' in df.columns:
df_min_date = df['date'].min()
df_max_date = df['date'].max()
min_date = max(min_date, df_min_date)
max_date = min(max_date, df_max_date)
return min_date, max_date
dataframes = [df1, df2, df3, df4, df5, df6, df7, df8, df9, df10, cle_inf, cle_eva, cle_pre]
min_date, max_date = find_min_max_dates(dataframes)
print(min_date)
print(max_date)
def reduce_dataframe_by_date_range(df, min_date, max_date):
mask = (df['date'] >= min_date) & (df['date'] <= max_date)
reduced_df = df.loc[mask].copy()
reduced_df.reset_index(drop=True, inplace=True)
return reduced_df
ao_df = reduce_dataframe_by_date_range(df1, min_date, max_date)
epo_df = reduce_dataframe_by_date_range(df2, min_date, max_date)
nao_df = reduce_dataframe_by_date_range(df3, min_date, max_date)
nino12_df = reduce_dataframe_by_date_range(df4, min_date, max_date)
nino3_df = reduce_dataframe_by_date_range(df5, min_date, max_date)
nino34_df = reduce_dataframe_by_date_range(df6, min_date, max_date)
nino4_df = reduce_dataframe_by_date_range(df7, min_date, max_date)
pna_df = reduce_dataframe_by_date_range(df8, min_date, max_date)
soi_df = reduce_dataframe_by_date_range(df9, min_date, max_date)
wpo_df = reduce_dataframe_by_date_range(df10, min_date, max_date)
cle_inflow_df = reduce_dataframe_by_date_range(cle_inf, min_date, max_date)
cle_evaporation_df = reduce_dataframe_by_date_range(cle_eva, min_date, max_date)
cle_precipitation_df = reduce_dataframe_by_date_range(cle_pre, min_date, max_date)
ao = pd.DataFrame()
epo = pd.DataFrame()
nao = pd.DataFrame()
nino12 = pd.DataFrame()
nino3 = pd.DataFrame()
nino34 = pd.DataFrame()
nino4 = pd.DataFrame()
pna = pd.DataFrame()
soi = pd.DataFrame()
wpo = pd.DataFrame()
cle_inflow = pd.DataFrame()
cle_evaporation = pd.DataFrame()
cle_precipitation = pd.DataFrame()
seasonality = pd.DataFrame()
ao['VALUE'] = ao_df.iloc[:,0]
epo['VALUE'] = epo_df.iloc[:,0]
nao['VALUE'] = nao_df.iloc[:,0]
nino12['VALUE'] = nino12_df.iloc[:,0]
nino3['VALUE'] = nino3_df.iloc[:,0]
nino34['VALUE'] = nino34_df.iloc[:,0]
nino4['VALUE'] = nino4_df.iloc[:,0]
pna['VALUE'] = pna_df.iloc[:,0]
soi['VALUE'] = soi_df['SOI'].copy()
wpo['VALUE'] = wpo_df.iloc[:,0]
cle_inflow['VALUE'] = cle_inflow_df['VALUE'].copy()
cle_evaporation['VALUE'] = cle_evaporation_df['VALUE'].copy()
cle_precipitation['VALUE'] = cle_precipitation_df['VALUE'].copy()
seasonality['VALUE'] = epo_df['date'].dt.month
def converttofloat(s):
count = 0
s = str(s)
v = float(s.replace(',',''))
if v < 0:
count +=1
return np.nan
else:
return v
cle_inflow['VALUE'] = cle_inflow['VALUE'].apply(converttofloat)
seasonality['VALUE'] = seasonality['VALUE'].apply(converttofloat)
cle_inflow['VALUE'].interpolate(method='linear', inplace=True)
cle_evaporation['VALUE'].interpolate(method='linear', inplace = True)
cle_precipitation['VALUE'].interpolate(method = 'linear', inplace=True)
ao['VALUE'].interpolate(method = 'linear', inplace=True)
def cfstom3(value):
value = value*0.0283168466
return value
def inchestomm(value):
value = value*25.4
return value
cle_inflow['VALUE'] = cle_inflow['VALUE'].apply(cfstom3)
cle_evaporation['VALUE'] = cle_evaporation['VALUE'].apply(cfstom3)
cle_precipitation['VALUE'] = cle_precipitation['VALUE'].apply(inchestomm)
def generate_lagged_data(data, data_name, time_lag):
data_lag = pd.DataFrame()
for i in range(time_lag-1,0,-1):
data_lag[data_name+'-'+str(i)] = data.shift(i).values[:,0]
data_lag[data_name] = data.values[:,0]
data_lag = data_lag[time_lag:]
return data_lag
ao_lag = generate_lagged_data(ao, 'ao_t', 3)
epo_lag = generate_lagged_data(epo, 'epo_t', 3)
nao_lag = generate_lagged_data(nao, 'nao_t', 3)
nino12_lag = generate_lagged_data(nino12, 'nino12_t', 3)
nino3_lag = generate_lagged_data(nino3, 'nino3_t', 3)
nino34_lag = generate_lagged_data(nino34, 'nino34_t', 3)
nino4_lag = generate_lagged_data(nino4, 'nino4_t', 3)
pna_lag = generate_lagged_data(pna, 'pna_t', 3)
soi_lag = generate_lagged_data(soi, 'soi_t', 3)
wpo_lag = generate_lagged_data(wpo, 'wpo_t', 3)
seasonality_lag = generate_lagged_data(seasonality, 'seasonality_t', 3)
cle_evaporation_lag = generate_lagged_data(cle_evaporation, 'evaporation_t', 3)
cle_precipitation_lag = generate_lagged_data(cle_precipitation, 'precipitation_t', 3)
inflow = pd.DataFrame()
cle_inflow_lag = pd.DataFrame()
for i in range(2,0,-1):
cle_inflow_lag['inflow_t-'+str(i)] = cle_inflow.shift(i).values[:,0]
cle_inflow_lag = cle_inflow_lag[3:]
inflow['VALUE'] = cle_inflow.values[:,0]
inflow = inflow[3:]
df_list = [ao_lag,epo_lag,nao_lag,nino12_lag,nino3_lag,nino34_lag,nino4_lag,pna_lag,soi_lag,wpo_lag,seasonality_lag,cle_evaporation_lag,cle_precipitation_lag,cle_inflow_lag]
X = pd.concat(df_list,axis=1)
y = inflow.copy()
scaler = StandardScaler()
X_scaler = scaler.fit_transform(X)
pca = PCA().fit(X_scaler)
plt.plot(np.cumsum(pca.explained_variance_ratio_))
plt.xlabel('Number of components')
plt.ylabel('Cumulative explained variance')
plt.title('PCA for CLE Resercoir')
plt.grid()
plt.show()
pca = PCA(n_components=15)
X_pca = pca.fit_transform(X_scaler)
def correlation(s, o):
if s.size == 0:
corr = np.NaN
else:
corr = np.corrcoef(o, s)[0,1]
return corr
def NS(s, o):
return 1 - np.sum((s-o)**2)/np.sum((o-np.mean(o))**2)
def KGE(s, o):
cc = correlation(s,o)
alpha = np.std(s)/np.std(o)
beta = np.sum(s)/np.sum(o)
kge = 1- np.sqrt( (cc-1)**2 + (alpha-1)**2 + (beta-1)**2 )
return kge
from scipy.stats import pearsonr
def CORR(s,o):
corr, _ = pearsonr(o,s)
return corr
def Absolute_Percentage_Error(s,o):
return (np.sum(np.abs(s-o))/np.sum(o))*100
from keras.layers import Dense
from keras.layers import LSTM
from keras.callbacks import EarlyStopping
from keras.models import Sequential
from sklearn.model_selection import train_test_split
x_lstm = X_pca.reshape(X_pca.shape[0],1,X_pca.shape[1])
stop_noimprovement = EarlyStopping(monitor='loss', patience=10)
model = Sequential()
model.add(LSTM(units=1024, activation='relu', return_sequences=True, input_shape=(1, x_lstm.shape[2])))
model.add(LSTM(units=1024, activation='relu', return_sequences=True))
model.add(Dense(units=1,activation='relu'))
model.compile(loss='mean_squared_error', optimizer='adam')
kf = KFold(n_splits=15, shuffle=False)
score_30_rmse = []
score_30_corr = []
score_30_r2 = []
score_30_mae = []
score_30_apb = []
r_lstm_1 = []
p_lstm_1 = []
for train_index, test_index in tqdm(kf.split(x_lstm),total=kf.get_n_splits(),desc="k-fold"):
X_train, X_test, y_train, y_test = x_lstm[train_index], x_lstm[test_index], y.iloc[train_index], y.iloc[test_index]
lstm_history = model.fit(X_train, y_train, epochs=30, batch_size=16, validation_data=(X_test, y_test), callbacks=[stop_noimprovement], shuffle=False,verbose=False)
yht = model.predict(X_test)
yhat= []
for i in yht:
if i[0][0] < 0:
yhat.append(0)
else:
yhat.append(i[0][0])
r_lstm_1.extend(list(y_test['VALUE']))
p_lstm_1.extend(yhat)
yhat = np.array(yhat)
y_test = np.array(list(y_test['VALUE']))
score_30_rmse.append(math.sqrt(mean_squared_error(y_test, yhat)))
score_30_corr.append(correlation(y_test,yhat))
score_30_r2.append(r2_score(y_test,yhat))
score_30_mae.append(mean_absolute_error(y_test,yhat))
score_30_apb.append(Absolute_Percentage_Error(y_test,yhat))
lstm_30 = pd.DataFrame({'actual':r_lstm_1,'predict':p_lstm_1})
print('RMSE ',mean(score_30_rmse))
print('CORR ',mean(score_30_corr))
print('R2 ',mean(score_30_r2))
print('MAE ',mean(score_30_mae))
print('APB ',mean(score_30_apb))