How to transform an MLP neural network into an RNN?

I have an MLP rating to predict the prices of the stock market prediction [0:] and the sentiment score on the 'sentiment_score' items. Now I want to try an RNN to do the same. However, I cook during the adjustment.

The MLP classifier was:

from sklearn.neural_network import MLPClassifier
from datetime import datetime, timedelta

years = [2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016]
prediction_list = []
for year in years:
    # Splitting the training and testing data
    train_start_date = str(year) + '-01-01'
    train_end_date = str(year) + '-10-31'
    test_start_date = str(year) + '-11-01'
    test_end_date = str(year) + '-12-31'
    train = df.ix[train_start_date : train_end_date]
    test = df.ix[test_start_date:test_end_date]

    # Calculating the sentiment score
    sentiment_score_list = []
    for date, row in train.T.iteritems():
        sentiment_score = np.asarray([df.loc[date, 'compound'],df.loc[date, 'neg'],df.loc[date, 'neu'],df.loc[date, 'pos']])
        #sentiment_score = np.asarray([df.loc[date, 'neg'],df.loc[date, 'pos']])
        sentiment_score_list.append(sentiment_score)
    numpy_df_train = np.asarray(sentiment_score_list)
    sentiment_score_list = []
    for date, row in test.T.iteritems():
        sentiment_score = np.asarray([df.loc[date, 'compound'],df.loc[date, 'neg'],df.loc[date, 'neu'],df.loc[date, 'pos']])
        #sentiment_score = np.asarray([df.loc[date, 'neg'],df.loc[date, 'pos']])
        sentiment_score_list.append(sentiment_score)
    numpy_df_test = np.asarray(sentiment_score_list)

    # Generating models
    mlpc = MLPClassifier(hidden_layer_sizes=(100, 200, 100), activation='relu', 
                         solver='lbfgs', alpha=0.005, learning_rate_init = 0.001, shuffle=False) # span = 20 # best 1
    mlpc.fit(numpy_df_train, train['prices'])   
    prediction = mlpc.predict(numpy_df_test)

    prediction_list.append(prediction)
    #print train_start_date + ' ' + train_end_date + ' ' + test_start_date + ' ' + test_end_date
    idx = pd.date_range(test_start_date, test_end_date)
    #print year
    predictions_df_list = pd.DataFrame(data=prediction[0:], index = idx, columns=['prices'])

    difference_test_predicted_prices = offset_value(test_start_date, test, predictions_df_list)
    # Adding offset to all the advpredictions_df price values
    predictions_df_list['prices'] = predictions_df_list['prices'] + difference_test_predicted_prices
    predictions_df_list

    # Smoothing the plot
    span = 20
    # modification for ewma
    sma = predictions_df_list["prices"].rolling(window=span, min_periods=span).mean()[:span]
    rest = predictions_df_list["prices"][span:]
    df_ewm = pd.concat([sma, rest]).ewm(span=span, adjust=False).mean()
    predictions_df_list['ewma']= df_ewm
    # end of modifications
    predictions_df_list['actual_value'] = test['prices']
    # modification for actual_value_ewma
    sma = predictions_df_list["actual_value"].rolling(window=span, min_periods=span).mean()[:span]
    rest = predictions_df_list["actual_value"][span:]
    df_actual_value_ewma = pd.concat([sma, rest]).ewm(span=span, adjust=False).mean()
    #predictions_df_list['actual_value_ewma'] = pd.DataFrame.ewm(predictions_df_list["actual_value"], span=20)
    predictions_df_list['actual_value_ewma'] = df_actual_value_ewma
    # Changing column names
    predictions_df_list.columns = ['predicted_price', 'average_predicted_price', 'actual_price', 'average_actual_price']
    predictions_df_list.plot()
    predictions_df_list_average = predictions_df_list[['average_predicted_price', 'average_actual_price']]
    predictions_df_list_average.plot()

#     predictions_df_list.show()

It works well! Well, you do not have the offset_value () but it's supposed to work. So I have this:

import numpy
import matplotlib.pyplot as plt
import pandas
import math

from keras.models import Sequential
from keras.layers import Dense
from keras.layers import LSTM
from sklearn.preprocessing import MinMaxScaler
from sklearn.metrics import mean_squared_error

import numpy as np
from keras.models import Sequential
from keras.layers import Dense, Dropout  

# load the dataset
years = [2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016]
prediction_list = []
for year in years:
    # Splitting the training and testing data
    train_start_date = str(year) + '-01-01'
    train_end_date = str(year) + '-10-31'
    test_start_date = str(year) + '-11-01'
    test_end_date = str(year) + '-12-31'
    train = df.ix[train_start_date : train_end_date]
    test = df.ix[test_start_date:test_end_date]

    # Calculating the sentiment score
    sentiment_score_list = []
    for date, row in train.T.iteritems():
        sentiment_score = np.asarray([df.loc[date, 'compound'],df.loc[date, 'neg'],df.loc[date, 'neu'],df.loc[date, 'pos']])
        #sentiment_score = np.asarray([df.loc[date, 'neg'],df.loc[date, 'pos']])
        sentiment_score_list.append(sentiment_score)
    numpy_df_train = np.asarray(sentiment_score_list)
    sentiment_score_list = []
    for date, row in test.T.iteritems():
        sentiment_score = np.asarray([df.loc[date, 'compound'],df.loc[date, 'neg'],df.loc[date, 'neu'],df.loc[date, 'pos']])
        #sentiment_score = np.asarray([df.loc[date, 'neg'],df.loc[date, 'pos']])
        sentiment_score_list.append(sentiment_score)
    numpy_df_test = np.asarray(sentiment_score_list)



    # create and fit the LSTM network
    model = Sequential()
    model.add(Dense(64, input_dim=4, activation='relu'))
    model.add(Dropout(0.5))
    model.add(Dense(64, activation='relu'))
    model.add(Dropout(0.5))
    model.add(Dense(1, activation='sigmoid'))
    model.compile(loss='binary_crossentropy',
              optimizer='rmsprop',
              metrics=['accuracy'])

    model.fit(numpy_df_test, y_train,
              epochs=20,
              batch_size=128)
    score = model.evaluate(x_test, y_test, batch_size=128)

and I have the following error:

  

ValueError: the input matrices must have the same number of   samples that the destination matrices. 61 samples of   input and 2922 target samples.