Executing Python script in MATLAB?

Is there any way to run a full python script ? I am not looking to just call a function written in python.

Answer 1

[Yongjian Feng]

What is the difference between a full python script and a function written in python? They are the same in my opinion.

[Abhishek Sharma]

import pandas as pd

import random

import numpy as np

import matplotlib.pyplot as plt

from scipy import io

import pickle as pkl

import keras

from keras.models import Sequential, Model, load_model

from keras.layers import LSTM, Dense, RepeatVector, TimeDistributed, Input, BatchNormalization, \

multiply, concatenate, Flatten, Activation, dot

from keras.optimizers import Adam

from keras.utils import plot_model

# from keras.callbacks import EarlyStopping

import pydot as pyd

from keras.utils.vis_utils import plot_model, model_to_dot

keras.utils.vis_utils.pydot = pyd

import numpy as np

import math

import tensorflow as tf

from pandas import DataFrame

from pandas import concat

from pandas import read_csv

from keras.models import Sequential

from keras.layers import Dense, TimeDistributed

from keras.layers import LSTM

from sklearn.preprocessing import MinMaxScaler

from sklearn.metrics import mean_squared_error

from tensorflow.python.keras.callbacks import TensorBoard

import datetime

keras=tf.keras

import scipy.io

from scipy import io

# Get x values of the sine wave

time= np.arange(0, 200, 0.01);

mat = scipy.io.loadmat('y.mat')

amplitude=mat['Y']

# Amplitude of the sine wave is sine of a variable like time

# amplitude= time*0.002+0.02

# amplitude=np.sin(2*math.pi*time)+np.sin(3*2*math.pi*time)+np.sin(5* 2*math.pi*time)

# noise = np. random. normal(0, .2, amplitude_. shape)

# amplitude=amplitude_+noise

# amplitude = scipy.io.loadmat('data.mat')

# amplitude=amplitude['data']

amplitude=amplitude/np.max(amplitude)

df=pd.DataFrame(amplitude)

df_train=amplitude[0:12000]

df_val=amplitude[12000:15000]

df_test=amplitude[15000:20000]

df_train=pd.DataFrame(df_train)

df_val=pd.DataFrame(df_val)

df_test=pd.DataFrame(df_test)

def series_to_supervised(data, n_in=1, n_out=1, dropnan=True):

n_vars = 1 if type(data) is list else data.shape[1]

df = DataFrame(data)

cols, names = list(), list()

# input sequence (t-n, ... t-1)

for i in range(n_in, 0, -1):

cols.append(df.shift(i))

names += [('var%d(t-%d)' % (j+1, i)) for j in range(n_vars)]

# forecast sequence (t, t+1, ... t+n)

for i in range(0, n_out):

cols.append(df.shift(-i))

if i == 0:

names += [('var%d(t)' % (j+1)) for j in range(n_vars)]

else:

names += [('var%d(t+%d)' % (j+1, i)) for j in range(n_vars)]

# put it all together

agg = concat(cols, axis=1)

agg.columns = names

# drop rows with NaN values

if dropnan:

agg.dropna(inplace=True)

return agg

n_in=100

gap=100 # number of steps after which forecast is needed

n_out=100+gap # horizon of forecast

reframed_train=series_to_supervised(df_train,n_in,n_out)

reframed_val=series_to_supervised(df_val,n_in,n_out)

reframed_test=series_to_supervised(df_test,n_in,n_out)

# reframed_train=reframed_train.drop(['var(t)])

for i in range(1,gap):

reframed_train=reframed_train.drop(['var1(t+%s)' %(i)],axis=1)

reframed_train=reframed_train.drop(['var1(t)'],axis=1)

# for i in range(1,n_out):

# reframed_train=reframed_train.drop(['var2(t+%s)' %(i)],axis=1)

# reframed_train=reframed_train.drop(['var2(t)'],axis=1)

for i in range(1,gap):

reframed_val=reframed_val.drop(['var1(t+%s)' %(i)],axis=1)

reframed_val=reframed_val.drop(['var1(t)'],axis=1)

# for i in range(1,n_out):

# reframed_val=reframed_val.drop(['var2(t+%s)' %(i)],axis=1)

# reframed_val=reframed_val.drop(['var2(t)'],axis=1)

for i in range(1,gap):

reframed_test=reframed_test.drop(['var1(t+%s)' %(i)],axis=1)

reframed_test=reframed_test.drop(['var1(t)'],axis=1)

# for i in range(1,n_out):

# reframed_test=reframed_test.drop(['var2(t+%s)' %(i)],axis=1)

# reframed_test=reframed_test.drop(['var2(t)'],axis=1)

reframed_train=reframed_train.values

reframed_val=reframed_val.values

reframed_test=reframed_test.values

train_X=reframed_train[:,0:(df.shape[1]*n_in)]

train_Y=reframed_train[:,df.shape[1]*n_in:(df.shape[1]*n_in)+df.shape[1]*n_out]

val_X=reframed_val[:,0:(df.shape[1]*n_in)]

val_Y=reframed_val[:,df.shape[1]*n_in:(df.shape[1]*n_in)+df.shape[1]*n_out]

test_X=reframed_test[:,0:(df.shape[1]*n_in)]

test_Y=reframed_test[:,df.shape[1]*n_in:(df.shape[1]*n_in)+df.shape[1]*n_out]

train_X = train_X.reshape((train_X.shape[0],n_in, df.shape[1]))

train_Y = train_Y.reshape((train_Y.shape[0],n_out-gap,1))

val_X = val_X.reshape((val_X.shape[0],n_in, df.shape[1]))

val_Y = val_Y.reshape((val_Y.shape[0],n_out-gap,1))

test_X=test_X.reshape((test_X.shape[0],n_in, df.shape[1]))

test_Y = test_Y.reshape((test_Y.shape[0],n_out-gap,1))

# def create_model():

# return tf.keras.models.Sequential([

# tf.keras.layers.Bidirectional(LSTM(12 8,return_sequences=False,stateful=False,batch_input_shape=(None,n_in,1))),

# # tf.keras.layers.Bidirectional(LSTM(50,return_sequences=True)),

# tf.keras.layers.Dense(n_out-gap)])

# #model.add(Dense(1))

# model = create_model()

# model.compile(loss=keras.losses.Huber(), optimizer='adam',metrics=["mae"])

# early_stopping = keras.callbacks.EarlyStopping(patience=10)

# history=model.fit(train_X, train_Y,validation_data=(val_X, val_Y),batch_size=32,epochs=1,callbacks=[early_stopping],shuffle=True)

# history=model.fit(train_X, train_Y,validation_data=(val_X, val_Y),batch_size=1,epochs=50,shuffle=False)

# # model.save('sinus.h5')

# m=model.predict(test_X)

# m = m.reshape((m.shape[0],n_out-gap,1))

# io.savemat("forecast_lin.mat",{"forecast_lin": m})

# io.savemat("true.mat",{"true": test_Y})

# error=np.abs(test_Y-m)

# mape=np.mean(np.abs((test_Y - m) / m)) * 100

# print('MAPE is: %.2f percent' % (mape))

[Abhishek Sharma]

I am looking to run this in matlab

[Yongjian Feng]

It seems to me that you actually want to run a ML script that calls other ML packages.

This can be done. You might need to play with the env var PYTHONPATH from matlab to point to the installed python packages.

According to experience of the other users who did this successfully, python virtualenv could cause some problems. If you don't have to use virtualenv, avoid it.

[Abhishek Sharma]

Can you please share the method or any link to it.

[Yongjian Feng]

There are several MATLAB Answers related to this:

https://www.mathworks.com/matlabcentral/answers/880623-annaconda-packages-not-loadable-in-matlab

There is another one regarding python virtualenv on Mac. I have trouble to find.

My suggestion, give it a try. If there is any error message, post it and we can look at it together.

[Abhishek Sharma]

I tried i but it gave following error:

Traceback (most recent call last):

File "ez_train_stlf.py", line 9, in <module>

import pandas as pd

ModuleNotFoundError: No module named 'pandas'

[Abhishek Sharma]

I think it's not able to import the modules.

[Yongjian Feng]

1. Is pandas installed properly in python? Can you use python to import it without problems?
2. Figure out where from it is imported, then inside matlab, call setenv (https://www.mathworks.com/help/matlab/ref/setenv.html) to set PYTHONPATH to add that location to it.

If you don't use virtualenv, this can be done easily.