Homec4science
Diffusion solar_potential (rc_energy_potential) Browse

solarPOT/Solar/HPELM_RC.py
30ee90157a72rc_energy_potential

Last Change
Raw File

HPELM_RC.py
View Options

import sys
import numpy as np
import matplotlib.pyplot as plt
from matplotlib.colors import LinearSegmentedColormap
import pandas as pd
import xarray as xr
import os
import time
import hpelm
import util
from ds import Dataset
from tables import open_file, Atom, Filters
print("Executing", sys.argv[0])
# ### Initialise reader
# In[3]:
# data_path = os.path.abspath("D:/Meteo_data/Meteoswiss_satellite")
#data_path = os.path.abspath("/home/castello/work/energy-potential/Solar/data/meteo")
#modelname = "2012-2012_grid100_SIS"
#queryname = 'grid1600'
data_path = os.path.abspath(sys.argv[1])
modelname = sys.argv[2]
queryname = sys.argv[3]
nodes = int(sys.argv[4])
# In[4]:
my_ds = Dataset(data_path, modelname, queryname)
# In[5]:
print(my_ds.train_X)
print(my_ds.val_X)
print(my_ds.test_X)
print(my_ds.query_X)
# In[6]:
nf = my_ds.feature_norm.m
nt = my_ds.target_norm.m
# In[7]:
my_ds.get_matrices(['train', 'test', 'val', 'query'])
# In[8]:
modelfile = os.path.join(my_ds.model_path, 'model.hdf5')
# ## HPELM
# In[9]:
model = hpelm.HPELM(nf, nt)
print(model)
# In[13]:
model.add_neurons(nodes, 'sigm')
print("number of nodes =", nodes)
#model.add_neurons(10, 'lin')
# In[14]:
timer = time.process_time()
model.train(my_ds.train_X, my_ds.train_T, 'V', Xv=my_ds.val_X, Tv=my_ds.val_T)
print("TRAINING:Executed in %.2f seconds" %(time.process_time()-timer))
# In[15]:
model.nnet.get_neurons()
# In[17]:
# TRAINING ERROR:
timer = time.clock()
model.predict(my_ds.train_X, my_ds.train_Y)
print(model.error(my_ds.train_Y, my_ds.train_T))
print("TRAINING ERROR: Executed in %.2f seconds" %(time.clock()-timer))
# In[18]:
# VALIDATION ERROR:
timer = time.clock()
model.predict(my_ds.val_X, my_ds.val_Y)
print(model.error(my_ds.val_Y, my_ds.val_T))
print("VALIDATION ERROR: Executed in %.2f seconds" %(time.clock()-timer))
# In[19]:
# TESTING ERROR:
timer = time.clock()
model.predict(my_ds.val_X, my_ds.test_Y)
print(model.error(my_ds.test_Y, my_ds.test_T))
print("TESTING ERROR: Executed in %.2f seconds" %(time.clock()-timer))
# In[24]:
timer = time.clock()
query_Y=model.predict(my_ds.query_X)
print("QUERY PREDICTIONS: Executed in %.2f seconds" %(time.clock()-timer))
# In[37]:
h5 = open_file(my_ds.query_X, "r")
for node in h5.walk_nodes():
pass # find a node with whatever name
X = node[:]
h5.close()
print("==== Query features === ")
print(X)
print("==== Query targets === ")
print(query_Y)
c4science · Help