prediction.py

import streamlit as st
import pandas as pd
import numpy as np
import nltk
import tensorflow as tf
from nltk.corpus import stopwords
import re
from nltk.tokenize import word_tokenize
from nltk.stem import WordNetLemmatizer
from keras.models import load_model




# Load model tf fromat.
new_model = tf.keras.models.load_model('lstm1_model')




def run():

  with st.form(key='Review & Rating'):

      Review = st.text_input('Review your stay', value='')
      Rating  = st.selectbox('Rate us', (1,2,3,4,5), index=3, help='1 = very bad , 5 = very nice')
      st.markdown('---')
           
      submitted = st.form_submit_button('Predict')

  data_inf = {
    'Review':Review,
    'Rating': Rating,
  }

  data_inf = pd.DataFrame([data_inf])
    
  st.dataframe(data_inf)
      
  # Define Stopwords

  ## Load Stopwords from NLTK
  nltk.download('stopwords')
  nltk.download('punkt')

  stw_en = stopwords.words("english")

  ## Create A New Stopwords
  new_stw = [ 'hotel', 'room','rooms','good','day','resort','night','restaurant','people','time', "n't", 'got' ,
   'staff',
   'stay',
   'location',
   'service',
   'stayed',
   'beach',
   'breakfast',
   'clean',
   'food',
   'place',
   'pool',
   'like',
   'really',
   'bed',
   'area',
    'bar',
   'small',
   'walk',
   'little',
   'bathroom',
   'trip',
   'floor',
   'minute',
   'water',
   'lot',
   'great',
   'nice',
   'went',
   'thing',
   'problem',
   'want',
   'drink',
   'way',
   'get',
   'go',
   'say'
   ]

  ## Merge Stopwords
  stw_en = stw_en + new_stw
  stw_en = list(set(stw_en))


  # Membuat Function untuk preprocessing kata dalam dataframe


  def text_proses(teks):
    # Mengubah Teks ke Lowercase
    teks = teks.lower()

    # Menghilangkan Mention
    teks = re.sub("@[A-Za-z0-9_]+", " ", teks)

    # Menghilangkan Hashtag
    teks = re.sub("#[A-Za-z0-9_]+", " ", teks)

    # Menghilangkan \n
    teks = re.sub(r"\\n", " ",teks)

    # Menghilangkan Whitespace
    teks = teks.strip()


    # Menghilangkan Link
    teks = re.sub(r"http\S+", " ", teks)
    teks = re.sub(r"www.\S+", " ", teks)

    # Menghilangkan yang Bukan Huruf seperti Emoji, Simbol Matematika (seperti μ), dst
    teks = re.sub("[^A-Za-z\s']", " ", teks)

    # Melakukan Tokenisasi
    tokens = word_tokenize(teks)

    # Menghilangkan Stopwords
    teks = ' '.join([word for word in tokens if word not in stw_en])

    return teks

  # Function lemmatizer
  def lemmatize_text(text):
    sentence = []
    for word in text.split():
      lemmatizer = WordNetLemmatizer()
      sentence.append(lemmatizer.lemmatize(word, 'v'))
    return ' '.join(sentence)


  # Mengaplikasikan Semua Teknik Preprocessing ke dalam Semua Documents

  data_inf['text_processed'] = data_inf['Review'].apply(text_proses)
  data_inf

  # lemmatize review
  nltk.download('wordnet')
  data_inf['text_processed'] = data_inf['text_processed'].apply(lemmatize_text)
  data_inf

  inf = data_inf['text_processed']



  if submitted:  
      # Predict using model ann
      y_pred = new_model.predict(inf)
      y_pred_conv= np.where(y_pred >= 0.5, 1, 0)
      y_pred_df = pd.DataFrame(y_pred_conv, columns=['0', '1', '2'])
      y_pred_final=y_pred_df.idxmax(1).astype(int)

      if y_pred_final.any() == 2:
            st.write('## Dude, your guest gave Positive feedback')
      if y_pred_final.any() == 1:
            st.write('## Dude, your guest gave Neutral feedback')
      else:
            st.write('## Attention, your guest gave Negative feedback')

      
if __name__ == '__main__':
    run()