data = load_data.get_data()
input_data = data['input']
output_data = data['output']
train_in = input_data[train_idx]
train_out_truth = output_data[train_idx]
test_in = input_data[test_idx]
test_out_truth = output_data[test_idx]
#make our network and run dat dynamical boi
esn = pyESN.ESN(n_inputs=train_in.shape[1],
n_outputs=train_out_truth.shape[1],
n_reservoir=700,
spectral_radius=.8,
noise=.08,
silent=False,
sparsity=.92,
random_state=42)
print("training network...")
machine_training_out = esn.fit(train_in, train_out_truth, inspect=False)
print("done training network.")
for pred_col in range(1): #train_out_truth.shape[1] ) :
plt.title("training results for col " + str(pred_col))
plt.plot(machine_training_out[:, pred_col], label="machine pred")
plt.plot(train_out_truth[:, pred_col], label="truth")
plt.legend(loc='best')
plt.show()
data = load_data.get_data()
move_data = data['move_data'][:, 0:1]
neuron_data = data['neuron_data']
train_in = neuron_data[train_idx]
train_out_truth = move_data[train_idx]
test_in = neuron_data[test_idx]
test_out_truth = move_data[test_idx]
#make our network and run dat dynamical boi
esn = pyESN.ESN(n_inputs=train_in.shape[1],
n_outputs=train_out_truth.shape[1],
n_reservoir=1000,
spectral_radius=.3,
teacher_forcing=False,
noise=.001,
sparsity=.42,
random_state=42)
print("training network...")
machine_training_out = esn.fit(train_in, train_out_truth, inspect=False)
print("done training network.")
plt.title("training results")
plt.plot(machine_training_out[:, 0], label="machine pred")
plt.plot(train_out_truth[:, 0], label="truth")
plt.legend(loc='best')
plt.show()
#compute our err on trained output (should be small)