from google.colab import drive

drive.mount('./gdrive')

!ls

gdrive	sample_data

import os

os.chdir('./gdrive/My Drive/Google Colaboratory/Colab Notebooks/Data Science   Machine Learning/Beatles NLP/Data')

import pandas as pd
import numpy as np

import re
import matplotlib.pyplot as plt
import seaborn as sns
import string
import nltk
import warnings
warnings.filterwarnings('ignore', category=DeprecationWarning)

%matplotlib inline

!pip install tqdm --upgrade

from tqdm import tqdm, tqdm_notebook

# using tqdm to get a progress bar on long-loading processes
tqdm.pandas(tqdm_notebook)

# reading in the big Beatles data Excel file into a Pandas dataframe. 
# The Beatles data will continue everything we need, including song names, album names, songwriters, lyrics, and more

data = pd.read_excel('beatles_data.xlsx')

data.head()

list(data)

['Song', 'Album Debut', 'Songwriter(s)', 'Lead Vocal(s)', 'Year', 'Lyrics']

# We will remove any songs that have something for lyrics (aka non-instrumental songs)
data = data[data['Lyrics'].notnull()]

data['Songwriter(s)']

1                         Lennon
2                         Lennon
3                      McCartney
4                       Harrison
5      McCartney\n (with Lennon)
                 ...            
213                    McCartney
214           Lennon\n McCartney
215                    McCartney
216                       Lennon
217                       Lennon
Name: Songwriter(s), Length: 213, dtype: object

data[data['Songwriter(s)'] == 'Lennon']

import spacy
import en_core_web_sm

def preprocessing_songs(df):

  df['Lyric Line'] = df['Lyrics'].apply(lambda x: x.split('\n'))
  df = df.set_index(['Song', 'Album Debut', 'Songwriter(s)', 'Lead Vocal(s)', 'Year'])['Lyric Line'].apply(pd.Series).stack().reset_index().drop('level_5',1)
  df.rename(columns={0:'Lyric Line'}, inplace=True)

  print('Removing quotes from song...')
  df['Song'] = df['Song'].apply(lambda x: remove_song_quotes(x))

  print('Cleaning songwriter(s) column...')
  df['Songwriter(s)'] = df['Songwriter(s)'].apply(lambda x: condense_band_members(x))
  
  print('Cleaning lead vocals(s) column...')
  df['Lead Vocal(s)'] = df['Lead Vocal(s)'].apply(lambda x: condense_band_members(x))

  print('Counting the number of distinct lines in the lyrics.')
  df['Number Lines'] = df['Lyric Line'].apply(lambda x: number_lines(x))

  print('Counting the number of words in each song.')
  df['Number Words'] = df['Lyric Line'].apply(lambda x: number_words(x))

  print('Calculating the average number of words in each line.')
  df['Average Words Per Line'] = df.apply(lambda row: words_per_line(row['Number Lines'], row['Number Words']), axis=1)

  print('Counting the number of apostrophes in each song.')
  df['Number Apostrophes'] = df['Lyric Line'].apply(lambda x: apostrophe_count(x))

  print('Calculating the average number of words in each line.')
  df['Average Words Per Apostrophe'] = df.apply(lambda row: words_per_apostrophe(row['Number Words'], row['Number Apostrophes']), axis=1)

  print('Cleaning lyrics...')
  df['Cleaned Lyrics'] = clean_text(df, 'Lyric Line')

  return df

  

def pre_processing_songs_nlp(df):
  print('Extracting NLP features...')
  df['NLP Features'] = df['Cleaned Lyrics'].progress_apply(lambda x: nlp_pipeline(x))

  df = df.dropna()

  df['Tokens'] = df['NLP Features'].apply(lambda x: x['tokens'])

  df['Lemmas'] = df['NLP Features'].apply(lambda x: x['lemmas'])

  df['Lemmas Text'] = df['Lemmas'].apply(lambda x: ' '.join(x))

  df['Album Debut'] = df['Album Debut'].str.replace(r"\n", " ")
  
  # Replace two spaces with one space
  df['Album Debut'] = df['Album Debut'].str.replace(r"  ", " ")

  df = df.reset_index(drop=True)

  return df

  

def nlp_features(df):
  # Create dataframe of syntactic dependency relation counts 'dep'
    df_dep = df['NLP Features'].progress_apply(lambda x: nlp_dict_to_df(x, 'dep_counter'))
    df_dep = df_dep.fillna(value = 0)

    # Create dataframe of coarse-grained parts-of-speech counts 'pos'
    df_pos = df['NLP Features'].progress_apply(lambda x: nlp_dict_to_df(x, 'pos_counter'))
    df_pos = df_pos.fillna(value = 0)

    # Create dataframe of stop word counts 'stop'
    df_stop = df['NLP Features'].progress_apply(lambda x: nlp_dict_to_df(x, 'stop_counter'))
    df_stop = df_stop.fillna(value = 0)

    # Create dataframe of fine-grained parts-of-speech counts 'tag'
    df_tag = df['NLP Features'].progress_apply(lambda x: nlp_dict_to_df(x, 'tag_counter'))
    df_tag = df_tag.fillna(value = 0)

    # Combine all NLP dataframes
    df_spacy = pd.concat([df_stop, df_pos, df_tag, df_dep], axis=1)

    df_spacy = df_spacy.reset_index(drop=True)

    return df_spacy

def remove_song_quotes(song_row):
  # Removes any quotations around songs
  try:
    song_row = re.sub(r'"', '', song_row)
  except:
    pass

  return song_row

def condense_band_members(row):
  # Create new pandas df column for either the individual songwriter or group songwriters
  if row =='Harrison':
    return row
  elif row == 'Lennon':
    return row
  elif row == 'McCartney':
    return row
  elif row == 'Starr':
    return row
  elif row == 'Instrumental':
    return row
  else:
    return 'Multiple Beatles'

def number_lines(lyrics_row):
  # Counts the number of distinct lines in the lyrics
  try:
    number_lines = lyrics_row.count('\n') + 1
    return number_lines
  except:
    return 0

def number_words(lyrics_row):
  # Counts the number of words in each song
  try:
    list_of_words = lyrics_row.split()
    number_words = len(list_of_words)
    return number_words
  except:
    return 0

def words_per_line(number_lines_row, number_words_row):
    # Calculates the average words per line
    
    if number_words_row == 0:
        # Don't want to return np.nan, return 0
        return 0
    else:
        average_words_per_line = number_words_row / number_lines_row
        return average_words_per_line

def apostrophe_count(lyrics_row):
    # Counts the number of apostrophes
    
    try:
        number_apostrophes = lyrics_row.count("'")
        return number_apostrophes
    except:
        return 0

def words_per_apostrophe(number_words_row, number_apostrophes_row):
    # Calculates the average number of words used per apostrophe used
    
    if number_words_row == 0 or number_apostrophes_row == 0:
        # Don't want to return np.nan, return 0
        return 0
    else:
        average_words_per_apostrophe = number_words_row / number_apostrophes_row
        return average_words_per_apostrophe

def clean_text(df, df_column):
    """
    Removing line breaks, special characters.
    Setting all letters to lowercase -- EXCEPT for pronoun I, kept capitalized.
    """
    
    # Replacing line breaks with one space
    df['Cleaned Lyrics'] = df[df_column].str.replace(r"\n", " ")
    
    # Removing special characters
    df['Cleaned Lyrics'] = df['Cleaned Lyrics'].str.replace(r"[.,?!();:]", "")
    
    # Replace two spaces with one space
    df['Cleaned Lyrics'] = df['Cleaned Lyrics'].str.replace(r"  ", " ")
    
    # Lower-case all words
    df['Cleaned Lyrics'] = df['Cleaned Lyrics'].str.lower()
    
    # Upper-case the pronouns I -- this helps with Spacy Lemmatization
    df['Cleaned Lyrics'] = df['Cleaned Lyrics'].str.replace(r" i i ", " I I ")
    df['Cleaned Lyrics'] = df['Cleaned Lyrics'].str.replace(r" i ", " I ")
    df['Cleaned Lyrics'] = df['Cleaned Lyrics'].str.replace(r" i'", " I'")
    
    return df['Cleaned Lyrics']

def remove_song_quotes(song_row):
    """
    Removing any quotation marks around songs.
    """      
    
    # Check if the song names have quotes around them (mine did)
    try:
        song_row = re.sub(r'"', '', song_row)
    except:
        pass
    
    return song_row

def nlp_pipeline(text_document):
    """
    Takes in a string and runs it through Spacy's NLP pipeline consisting of a 
    Tokenizer, Tagger, Dependency Parser, Entity Recognizer, Text Categorizer.
    NLP features are then extracted from each Token's Spacy attributes.
    NLP features are then aggregated for the entire text_document and returned in a dictionary.
    
    Inputs:
        text_document (str): Text data. In this case it is Beatles lyrics.
    
    Output:
        dict: Aggregated NLP features.
    """
    
    # Create Spacy NLP pipeline
    nlp = en_core_web_sm.load() 
    
    # Check if text is not a string, if so return np.nan
    if type(text_document) != str:
        return np.nan
    
    # Run Spacy NLP pipeline on text_document, creates DOC object filled with tokens
    doc = nlp(text_document)

    # Tokenization
    tokens = [tok.orth_ for tok in doc]

    # Lemmatization
    lemmas = [tok.lemma_ for tok in doc]

    # Create counter dictionaries to collect counts of NLP features
    pos_counter = {}   # Coarse-grained part-of-speech
    tag_counter = {}   # Fine-grained part-of-speech.
    stop_counter = {}   # Stop word or not
    dep_counter = {}   # Syntactic dependency relation

    # Loop through each Token object contained in Doc object
    for token in doc:

        # Add token 'POS' to dictionary
        pos = "pos_" + token.pos_
        if pos in pos_counter.keys():
            pos_counter[pos] += 1
        else:
            pos_counter[pos] = 1

        # Add token 'TAG' to dictionary
        tag = "tag_" + token.tag_
        if tag in tag_counter.keys():
            tag_counter[tag] += 1
        else:
            tag_counter[tag] = 1

        # Add token 'STOP' to dictionary
        stop = "stop_" + str(token.is_stop)
        if stop in stop_counter.keys():
            stop_counter[stop] += 1
        else:
            stop_counter[stop] = 1

        # Add token 'DEP' to dictionary
        dep = "dep_" + token.dep_
        if dep in dep_counter.keys():
            dep_counter[dep] += 1
        else:
            dep_counter[dep] = 1

    # Combine NLP features into one dictionary
    nlp_dictionary = {'pos_counter' : pos_counter,
                      'tag_counter' : tag_counter,
                      'stop_counter' : stop_counter,
                      'dep_counter' : dep_counter,
                      'tokens' : tokens,
                      'lemmas' : lemmas}
    
    return nlp_dictionary

def nlp_dict_to_df(nlp_features, feature):
    """
    Take in nlp_features dictionary, outputs a pd.Series object 
    containing the NLP features.
    
    Inputs:
        nlp_features (dict): Aggregated NLP features.
    
    Outputs:
        nlp_series (pd.Series): NLP features for each song obtained using Spacy.
    """
    
    # Dep dictionary
    nlp_dict = nlp_features[feature]

    # Total number of entries in dep dictionary
    nlp_total = np.sum(list(nlp_dict.values()))

    # Calculate fraction of total for each NLP feature (Normalization)
    nlp_dict_fractions = {k: v / nlp_total for k, v in nlp_dict.items()}

    # Turn into a pandas Series to return
    nlp_series = pd.Series(nlp_dict_fractions)
    
    return nlp_series

# we have a new dataframe with cleaned lyrics
new_df = preprocessing_songs(data)

new_df.head()

# we will add all of the NLP features to the dataframe
new_df = pre_processing_songs_nlp(new_df)

spacy_df = nlp_features(new_df)

100%|██████████| 5664/5664 [00:02<00:00, 2174.94it/s]
100%|██████████| 5664/5664 [00:02<00:00, 2303.82it/s]
100%|██████████| 5664/5664 [00:02<00:00, 2196.54it/s]
100%|██████████| 5664/5664 [00:02<00:00, 2300.75it/s]

new_df.head()

spacy_df.head()

new_df.to_csv('data.csv')

spacy_df.to_csv('spacy.csv')

data = new_df
spacy = spacy_df

# Checking for class imbalance
data['Songwriter(s)'].value_counts()

Multiple Beatles    1927
Lennon              1629
McCartney           1488
Harrison             612
Starr                  8
Name: Songwriter(s), dtype: int64

data[data['Songwriter(s)'] == 'Starr']

data = data.drop([4495, 4496, 4497, 4498, 4499, 4500])

spacy = spacy.drop([4495, 4496, 4497, 4498, 4499, 4500])

data = data.drop([4501, 4502])
spacy = spacy.drop([4501, 4502])

data = data.reset_index(drop=True)
spacy = spacy.reset_index(drop=True)

data = data.drop("Unnamed: 0", axis=1)

data.head()

spacy = spacy.drop("Unnamed: 0", axis=1)

spacy.head()

data['Songwriter(s)'].value_counts()

Multiple Beatles    1927
Lennon              1629
McCartney           1488
Harrison             612
Name: Songwriter(s), dtype: int64

print(data.shape)
print(spacy.shape)

(5656, 16)
(5656, 104)

import itertools

from sklearn.model_selection import train_test_split
from sklearn.feature_extraction.text import CountVectorizer, TfidfVectorizer

data['Lemmas Text']

0       word be flow out like endless rain into a pape...
1       -PRON- slither while -PRON- pass -PRON- slip a...
2       pool of sorrow wave of joy be drift through -P...
3                               possess and caress -PRON-
4                                        jai guru deva om
                              ...                        
5651                             " love will find a way "
5652                        gather round all -PRON- clown
5653                           let -PRON- hear -PRON- say
5654         hey -PRON- have get to hide -PRON- love away
5655         hey -PRON- have get to hide -PRON- love away
Name: Lemmas Text, Length: 5656, dtype: object

# Define X and y
y = data['Songwriter(s)']
X = data['Lemmas Text']

# Train-test split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=.2, random_state = 444)

# Vocabulary of lemmas, use with CountVectorizer
vocabulary = set(itertools.chain.from_iterable(data['Lemmas']))

# Count vectorizer and the index-to-word mapping
count_vectorizer = CountVectorizer(vocabulary=vocabulary)

# Create bag-of-word embeddings using vectorizer
X_train_counts = count_vectorizer.fit_transform(X_train)
X_test_counts = count_vectorizer.transform(X_test)

# Convert to a pandas dataframe
X_train_counts = pd.DataFrame(X_train_counts.todense())
X_test_counts = pd.DataFrame(X_test_counts.todense())

# Create a mapping from index-to-word for the bag of words
index_to_word = {v:k for k,v in count_vectorizer.vocabulary_.items()}

X_train_counts

import matplotlib
import matplotlib.pyplot as plt
import matplotlib.patches as mpatches
from sklearn.decomposition import PCA, TruncatedSVD
from sklearn.preprocessing import LabelEncoder

le = LabelEncoder()
le.fit(y_train)
y_train_encoded = le.transform(y_train)

def plot_LSA(test_data, test_labels, plot=True):
    """
    This function first uses SK-Learn's truncated SVD (LSA) class to 
    transform the high dimensionality (number of columns) of the BoW 
    embedding down to 2 dimensions. Then the two dimensions are used
    to plot each song, colored by the song writer (class).
    
    Inputs:
        test_data (pd.DataFrame): BoW embeddings.
        test_labels (pd.Series): In this case the songwriter of each
        Beatles' song.
        plot (boolean): Whether or not to plot. Defaults to True.
    
    Outputs:
        None.
    """
    
    lsa = TruncatedSVD(n_components=2)
    lsa.fit(test_data)
    lsa_scores = lsa.transform(test_data)
    color_mapper = {label:idx for idx,label in enumerate(set(test_labels))}
    color_column = [color_mapper[label] for label in test_labels]
    colors = ['blue','green','purple', 'red']
    if plot:
        plt.scatter(lsa_scores[:,0], lsa_scores[:,1], s=8, alpha=.8, c=test_labels, cmap=matplotlib.colors.ListedColormap(colors))
        blue_patch = mpatches.Patch(color='blue', label='Harrison')
        green_patch = mpatches.Patch(color='green', label='Lennon')
        purple_patch = mpatches.Patch(color='purple', label='McCartney')
        orange_patch = mpatches.Patch(color='red', label='Multiple Beatles')
        plt.legend(handles=[blue_patch, green_patch, purple_patch, orange_patch], prop={'size': 18})
        plt.xlabel('Principal Component One')
        plt.ylabel('Principal Component Two')
        plt.rcParams["xtick.labelsize"] = 20
        plt.rcParams["ytick.labelsize"] = 20

# Run LSA on BoW embeddings and plot it
fig = plt.figure(figsize=(12, 12))
plot_LSA(X_train_counts, y_train_encoded)
plt.title('LSA: BoW', fontsize = 25)
plt.rcParams["xtick.labelsize"] = 18
plt.rcParams["ytick.labelsize"] = 18
plt.rcParams["axes.labelsize"] = 20
plt.xlim(-0.2, 3.2)
plt.ylim(-0.04, 0.06)
plt.show()

y = data['Songwriter(s)']
X = data['Lemmas Text']

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state = 444)

vocabulary = set(itertools.chain.from_iterable(data['Lemmas']))

tfidf_vectorizer = TfidfVectorizer(vocabulary=vocabulary)

X_train_tfidf = tfidf_vectorizer.fit_transform(X_train)
X_test_tfidf = tfidf_vectorizer.transform(X_test)

X_train_tfidf = pd.DataFrame(X_train_tfidf.todense())
X_test_tfidf = pd.DataFrame(X_test_tfidf.todense())

index_to_word = {v:k for k, v in count_vectorizer.vocabulary_.items()}

fig = plt.figure(figsize=(12,12))
plot_LSA(X_train_tfidf, y_train_encoded)
plt.title('LSA: BoW + TFIDF Transformation', fontsize= 25)
plt.rcParams["xtick.labelsize"] = 18
plt.rcParams["ytick.labelsize"] = 18
plt.rcParams["axes.labelsize"] = 20
plt.show()

y = data['Songwriter(s)']
X = spacy

le = LabelEncoder()
le.fit(y)
y_encoded = le.transform(y)

fig = plt.figure(figsize=(12, 12))
plot_LSA(X, y_encoded)
plt.title('LSA: Spacy NLP Features', fontsize = 25)
plt.rcParams["xtick.labelsize"] = 18
plt.rcParams["ytick.labelsize"] = 18
plt.rcParams["axes.labelsize"] = 20
plt.xlim(0.5, 1.2)
plt.show()

from sklearn.metrics import make_scorer, accuracy_score, classification_report, confusion_matrix
from sklearn.feature_extraction.text import TfidfTransformer
from sklearn.model_selection import GridSearchCV
from sklearn.pipeline import Pipeline

y = data['Songwriter(s)']
X = data['Cleaned Lyrics']

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2, random_state = 444)

vocabulary = set(itertools.chain.from_iterable(data['Cleaned Lyrics']))

count_vectorizer = CountVectorizer(vocabulary=vocabulary)

from sklearn.linear_model import LogisticRegression

!pip3 install pactools

from pactools.grid_search import GridSearchCVProgressBar

# Pipeline
pipeline = Pipeline([
    ('vect', CountVectorizer()),
    ('tfidf', TfidfTransformer()),
    ('clf', LogisticRegression()),])


# pipeline uses intermediate steps of CountVectorizer() and TfidfTransformer() before finally applying log regression

# This pipeline will be used as the estimator in the GridSearchCV


# Parameter Grid
param_grid = {
        'vect__max_df': (0.05, 0.075, 0.1, 0.15, 0.20, 0.25, 0.5, 0.75, 1.0),
        'vect__ngram_range': ((1, 1), (1, 2), (1, 3)),
        'vect__stop_words' : (None, 'english'),
        'tfidf__use_idf': (True, False),
        'tfidf__norm': ('l1', 'l2'),
        'clf__C': [0.001, 0.01, 0.1, 1, 10, 100, 1000]}


# Create the GridSearchCV object
#grid = GridSearchCV(pipeline, cv=10, n_jobs=-1, param_grid=param_grid, scoring=make_scorer(accuracy_score))
grid = GridSearchCVProgressBar(pipeline, cv=2, n_jobs=-1, param_grid=param_grid, scoring=make_scorer(accuracy_score))


# Run the grid search 
grid.fit(X_train, y_train)

# Predicts usings the best parameters of the grid search
y_pred = grid.predict(X_test)

# Accuracy Score
print('Accuracy Score:')
print(accuracy_score(y_test, y_pred))
print()

# Classification Report
print(classification_report(y_test, y_pred, digits=3))
print()

# Confusion Matrix
print('Confusion Matrix:')
print(confusion_matrix(y_test, y_pred))
print()

# Best parameters from the grid search
print('Best parameters set found on development set:')
print(grid.best_params_)
print()

 def get_most_important_features(vectorizer, model, n=5):
    """
    This function is used to find the most important ngrams (words) 
    for classifying each class (songwriter).
    
    Args:
        vectorizer (sklearn.feature_extraction.text.CountVectorizer):
        The vectorizer used to create the Bag of Words space matrix.
        model (sklearn.linear_model.logistic.LogisticRegression):
        The LR model trained on the dataset.
        n (int): Number of top words to return.
    
    Returns:
        important_words (dict): Dicitionary containing the most
        critical words used to predict each songwriter class.
    """
    
    # Create a mapping dictionary from words to index
    index_to_word = {v:k for k,v in vectorizer.vocabulary_.items()}
    
    # loop for each class
    important_words = {}
    for class_index in range(model.coef_.shape[0]):
        word_importances = [(el, index_to_word[i]) for i,el in enumerate(model.coef_[class_index])]
        sorted_coeff = sorted(word_importances, key = lambda x : x[0], reverse=True)
        tops = sorted(sorted_coeff[:n], key = lambda x : x[0])
        bottom = sorted_coeff[-n:]
        important_words[class_index] = {'tops':tops, 'bottom':bottom}
    
    return important_words

# Vectorize
vectorizer = CountVectorizer(max_df = 0.1, ngram_range = (1, 2))
X_train_counts = vectorizer.fit_transform(X_train).todense()
X_test_counts = vectorizer.transform(X_test).todense()

# Tfidf transformation
tfidf = TfidfTransformer(norm = 'l2', use_idf = False)
X_train_counts_tfidf = tfidf.fit_transform(X_train_counts).todense()
X_test_counts_tfidf = tfidf.fit_transform(X_test_counts).todense()

clf = LogisticRegression(C = 10)
clf.fit(X_train_counts_tfidf, y_train)

# Create predictions, using trained model
y_pred = clf.predict(X_test_counts_tfidf)

# Find the most important words used for classification
importance = get_most_important_features(vectorizer, clf, 10)

# Re-name the dictionary keys
importance['Harrison'] = importance.pop(0)
importance['Lennon'] = importance.pop(1)
importance['McCartney'] = importance.pop(2)
importance['Multiple Beatles'] = importance.pop(3)

# Print the most important words
importance

/usr/local/lib/python3.6/dist-packages/sklearn/linear_model/_logistic.py:940: ConvergenceWarning: lbfgs failed to converge (status=1):
STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.

Increase the number of iterations (max_iter) or scale the data as shown in:
    https://scikit-learn.org/stable/modules/preprocessing.html
Please also refer to the documentation for alternative solver options:
    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression
  extra_warning_msg=_LOGISTIC_SOLVER_CONVERGENCE_MSG)





{'Harrison': {'bottom': [(-2.045166065273923, 'cry'),
   (-2.0516036163996256, 'his'),
   (-2.074373830608004, 'down'),
   (-2.12649886774832, 'to make'),
   (-2.245646233845527, 'to be'),
   (-2.334401001365231, 'man'),
   (-2.603412111344155, 'yeah'),
   (-2.7788114268239608, 'better'),
   (-2.8376165191451923, 'but it'),
   (-3.245071662756538, 'he')],
  'tops': [(3.20668293232807, 'where you'),
   (3.2813107471829204, 'baby in'),
   (3.4526598680364975, 'it all'),
   (3.5652237899352883, 'really'),
   (3.654003225292855, 'too much'),
   (3.9902224848198466, 'got time'),
   (4.20846845653345, 'without'),
   (4.357034252936614, 'need you'),
   (4.579932179604316, 'to do'),
   (5.1964079560854515, 'me mine')]},
 'Lennon': {'bottom': [(-2.272389847767595, 'where'),
   (-2.2956971912605235, 'oh no'),
   (-2.3282633105894437, 'love you'),
   (-2.4086693661514875, 'here it'),
   (-2.4600020848978077, 'didn'),
   (-2.5905460013495665, 'know that'),
   (-2.614796732779806, 'my love'),
   (-2.838097648584486, 'without'),
   (-3.008428029862003, 'what you'),
   (-3.335549019342312, 'and she')],
  'tops': [(2.8503490406813254, 'oh now'),
   (2.9526071165781835, 'loser'),
   (3.0119376940263276, 'be alone'),
   (3.0762549486335216, 'you can'),
   (3.1254279080702903, 'nothing'),
   (3.1874394844050453, 'good night'),
   (3.205521442965275, 'well you'),
   (3.461479553947609, 'got to'),
   (3.769361331832148, 'but it'),
   (3.8967021106142603, 'get home')]},
 'McCartney': {'bottom': [(-2.1511610155645258, 'what'),
   (-2.241241980400747, 'their'),
   (-2.282678302634811, 'call'),
   (-2.4375112322430796, 'it true'),
   (-2.4586062152526638, 'me down'),
   (-2.5498405491478606, 'little girl'),
   (-2.6796571304298133, 'more'),
   (-2.7250424463530645, 'feel'),
   (-2.9858729016244396, 'think'),
   (-3.029644036108259, 'everything')],
  'tops': [(3.261969533610674, 'hello'),
   (3.359352795739651, 'want it'),
   (3.36567816522238, 'get back'),
   (3.3855530347876437, 'night before'),
   (3.4596632406431764, 'see yeah'),
   (3.5048001858749482, 'buy'),
   (3.5254689500809624, 'all my'),
   (3.6436609732156264, 'me back'),
   (3.7248591982508557, 'fast'),
   (4.011602595399383, 'and in')]},
 'Multiple Beatles': {'bottom': [(-2.380702445114838, 'can you'),
   (-2.4179787857734545, 'all you'),
   (-2.518133878680189, 'for'),
   (-2.5239313857912764, 'don'),
   (-2.662133994180265, 'me mine'),
   (-2.6709051555118153, 'around'),
   (-2.68177981063666, 'really'),
   (-2.8291554307864586, 'here'),
   (-2.892275500844382, 'still'),
   (-3.235839403441347, 'to do')],
  'tops': [(3.272339742398448, 'be your'),
   (3.3451528635041816, 'me do'),
   (3.356760928346278, 'anything'),
   (3.437673322366979, 'feeling'),
   (3.458280892290301, 'rocky'),
   (3.4692471969734586, 'know my'),
   (3.488929776316816, 'trouble'),
   (3.514775621809661, 'loves you'),
   (3.547161861254181, 'dance'),
   (3.581689596284666, 'better')]}}

from sklearn.naive_bayes import MultinomialNB

# Pipeline
pipeline = Pipeline([
    ('vect', CountVectorizer()),
    ('tfidf', TfidfTransformer()),
    ('clf', MultinomialNB())])

# Parameter Grid
param_grid = {
        'vect__max_df': (0.05, 0.075, 0.1, 0.15, 0.20, 0.25, 0.5, 0.75, 1.0),
        'vect__ngram_range': ((1, 1), (1, 2), (1, 3)),
        'vect__stop_words' : (None, 'english'),
        'tfidf__use_idf': (True, False),
        'tfidf__norm': ('l1', 'l2'),
        'clf__alpha': (0, 0.001, 0.01, 0.1, 0.2, 0.25, 0.5, 0.75, 1.0)}

# Create the GridSearchCV object
grid = GridSearchCV(pipeline, cv=10, n_jobs=-1, param_grid=param_grid, scoring=make_scorer(accuracy_score))

# Run the grid search 
grid.fit(X_train, y_train)

# Predict usings the best parameters of the grid search
y_pred = grid.predict(X_test)

# Accuracy Score
print('Accuracy Score:')
print(accuracy_score(y_test, y_pred))
print()

# Classification Report
print(classification_report(y_test, y_pred, digits=3))
print()

# Confusion Matrix
print('Confusion Matrix:')
print(confusion_matrix(y_test, y_pred))
print()

# Best parameters from the grid search
print('Best parameters set found on development set:')
print(grid.best_params_)
print()

# Vectorize
vectorizer = CountVectorizer(max_df = 0.1, ngram_range = (1, 2))
X_train_counts = vectorizer.fit_transform(X_train).todense()
X_test_counts = vectorizer.transform(X_test).todense()

# Tfidf transformation
tfidf = TfidfTransformer(norm = 'l2', use_idf = False)
X_train_counts_tfidf = tfidf.fit_transform(X_train_counts).todense()
X_test_counts_tfidf = tfidf.fit_transform(X_test_counts).todense()

nb = MultinomialNB()
nb.fit(X_train_counts_tfidf, y_train)

# Create predictions, using trained model
y_pred = nb.predict(X_test_counts_tfidf)

# Find the most important words used for classification
importance = get_most_important_features(vectorizer, nb, 10)

# Re-name the dictionary keys
importance['Harrison'] = importance.pop(0)
importance['Lennon'] = importance.pop(1)
importance['McCartney'] = importance.pop(2)
importance['Multiple Beatles'] = importance.pop(3)

# Print the most important words
importance

{'Harrison': {'bottom': [(-9.377282492934983, 'your trust'),
   (-9.377282492934983, 'your voice'),
   (-9.377282492934983, 'yours sincerely'),
   (-9.377282492934983, 'yours yet'),
   (-9.377282492934983, 'yourself home'),
   (-9.377282492934983, 'yourself in'),
   (-9.377282492934983, 'yourself to'),
   (-9.377282492934983, 'zapped'),
   (-9.377282492934983, 'zapped in'),
   (-9.377282492934983, 'zoo what')],
  'tops': [(-7.106033702676536, 'not'),
   (-7.099118495814187, 'll'),
   (-7.0610869856031915, 'that'),
   (-6.910337502354965, 'if'),
   (-6.870088757600826, 'be'),
   (-6.857382826987659, 'know'),
   (-6.831224065606969, 'love'),
   (-6.545435529591376, 'don'),
   (-6.285568876248815, 'all'),
   (-6.179713804302713, 'it')]},
 'Lennon': {'bottom': [(-9.557661329838217, 'your voice'),
   (-9.557661329838217, 'your window'),
   (-9.557661329838217, 'yours sincerely'),
   (-9.557661329838217, 'yourself'),
   (-9.557661329838217, 'yourself home'),
   (-9.557661329838217, 'yourself in'),
   (-9.557661329838217, 'yourself no'),
   (-9.557661329838217, 'yourself to'),
   (-9.557661329838217, 'zoo'),
   (-9.557661329838217, 'zoo what')],
  'tops': [(-6.234386539162086, 'in'),
   (-6.226543065041415, 'all'),
   (-6.226129185516754, 'that'),
   (-6.2054850694945145, 'know'),
   (-6.19591051876331, 'my'),
   (-6.188792385169995, 'can'),
   (-6.122542730040965, 'is'),
   (-6.01312671508577, 'love'),
   (-5.973525540057158, 'she'),
   (-5.684910114649124, 'it')]},
 'McCartney': {'bottom': [(-9.53302596278112, 'your vest'),
   (-9.53302596278112, 'your way'),
   (-9.53302596278112, 'your window'),
   (-9.53302596278112, 'yours yet'),
   (-9.53302596278112, 'yourself home'),
   (-9.53302596278112, 'yourself no'),
   (-9.53302596278112, 'zapped'),
   (-9.53302596278112, 'zapped in'),
   (-9.53302596278112, 'zoo'),
   (-9.53302596278112, 'zoo what')],
  'tops': [(-6.488101822072984, 'your'),
   (-6.48290034705385, 'down'),
   (-6.471984767206914, 'don'),
   (-6.446834475852382, 'my'),
   (-6.444107708474709, 'love'),
   (-6.4276420486337775, 'of'),
   (-6.372149255906219, 'she'),
   (-6.340993470408326, 'be'),
   (-6.0683605112704395, 'in'),
   (-5.89485844637437, 'it')]},
 'Multiple Beatles': {'bottom': [(-9.598712792303138, 'your taste'),
   (-9.598712792303138, 'your teddy'),
   (-9.598712792303138, 'your vest'),
   (-9.598712792303138, 'your voice'),
   (-9.598712792303138, 'yours'),
   (-9.598712792303138, 'yours sincerely'),
   (-9.598712792303138, 'yours yet'),
   (-9.598712792303138, 'yourself no'),
   (-9.598712792303138, 'zapped'),
   (-9.598712792303138, 'zapped in')],
  'tops': [(-6.170681412914195, 'yeah'),
   (-6.118398153346851, 'your'),
   (-6.016575553239925, 'all'),
   (-6.012901653314865, 'that'),
   (-5.998315932927394, 'know'),
   (-5.9692317567912685, 'be'),
   (-5.964634958768828, 'in'),
   (-5.9403216686676235, 'my'),
   (-5.881664089813732, 'it'),
   (-5.6590783376237415, 'love')]}}

from google.colab import files

import os, pandas as pd

os.chdir('./gdrive/My Drive/Google Colaboratory/Colab Notebooks/Data Science   Machine Learning/Beatles NLP/Data')

data = pd.read_csv('data.csv')

data.head()

data.rename(columns={'Unnamed: 0':'Line ID'}, inplace=True)

list(data)

['Line ID',
 'Song',
 'Album Debut',
 'Songwriter(s)',
 'Lead Vocal(s)',
 'Year',
 'Lyric Line',
 'Number Lines',
 'Number Words',
 'Average Words Per Line',
 'Number Apostrophes',
 'Average Words Per Apostrophe',
 'Cleaned Lyrics',
 'NLP Features',
 'Tokens',
 'Lemmas',
 'Lemmas Text']

data.head()

lyrics_data = data[['Line ID', 'Song', 'Album Debut', 'Songwriter(s)', 'Year', 'Cleaned Lyrics']]

lyrics_data.head()

lyrics_data = lyrics_data.set_index('Line ID')

lyrics_data.head()

# We will read in all of the cleaned lyrics into a single text file for training
with open('lyricsText.txt', 'w', encoding='utf-8') as filehandle:
  for listitem in lyrics_data['Cleaned Lyrics']:
    filehandle.write('%s\n' % listitem)

!pip install -q textgenrnn

from textgenrnn import textgenrnn

model_cfg = {
    'rnn_size': 128,
    'rnn_layers': 4,
    'rnn_bidirectional': True,
    'max_length': 40,
    'max_words': 10000,
    'dim_embeddings': 100,
    'word_level': False,
}

train_cfg = {
    'line_delimited': True,
    'num_epochs': 10, 
    'gen_epochs': 2,
    'batch_size': 1024,
    'train_size': 0.8,
    'dropout': 0.0,
    'max_gen_length': 300,
    'validation': True,
    'is_csv': False
}

text_file = 'lyricsText.txt'

model_name = '500nds_12Lrs_100epchs_Model'

model_name = '500nds_12Lrs_100epchs_Model'
textgen = textgenrnn(name=model_name)

train_function = textgen.train_from_file if train_cfg['line_delimited'] else textgen.train_from_largetext_file

train_function(
    file_path=text_file,
    new_model=True,
    num_epochs=train_cfg['num_epochs'],
    gen_epochs=train_cfg['gen_epochs'],
    batch_size=train_cfg['batch_size'],
    train_size=train_cfg['train_size'],
    dropout=train_cfg['dropout'],
    max_gen_length=train_cfg['max_gen_length'],
    validation=train_cfg['validation'],
    is_csv=train_cfg['is_csv'],
    rnn_layers=model_cfg['rnn_layers'],
    rnn_size=model_cfg['rnn_size'],
    rnn_bidirectional=model_cfg['rnn_bidirectional'],
    max_length=model_cfg['max_length'],
    dim_embeddings=model_cfg['dim_embeddings'],
    word_level=model_cfg['word_level'])

WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:541: The name tf.placeholder is deprecated. Please use tf.compat.v1.placeholder instead.

WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:4432: The name tf.random_uniform is deprecated. Please use tf.random.uniform instead.

WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:190: The name tf.get_default_session is deprecated. Please use tf.compat.v1.get_default_session instead.

WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:197: The name tf.ConfigProto is deprecated. Please use tf.compat.v1.ConfigProto instead.

WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:203: The name tf.Session is deprecated. Please use tf.compat.v1.Session instead.

WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:207: The name tf.global_variables is deprecated. Please use tf.compat.v1.global_variables instead.

WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:216: The name tf.is_variable_initialized is deprecated. Please use tf.compat.v1.is_variable_initialized instead.

WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:223: The name tf.variables_initializer is deprecated. Please use tf.compat.v1.variables_initializer instead.

WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:66: The name tf.get_default_graph is deprecated. Please use tf.compat.v1.get_default_graph instead.

WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/optimizers.py:793: The name tf.train.Optimizer is deprecated. Please use tf.compat.v1.train.Optimizer instead.

WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:3576: The name tf.log is deprecated. Please use tf.math.log instead.

5,663 texts collected.
Training new model w/ 4-layer, 128-cell Bidirectional LSTMs
Training on 134,882 character sequences.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow_core/python/ops/math_grad.py:1424: where (from tensorflow.python.ops.array_ops) is deprecated and will be removed in a future version.
Instructions for updating:
Use tf.where in 2.0, which has the same broadcast rule as np.where
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:1033: The name tf.assign_add is deprecated. Please use tf.compat.v1.assign_add instead.

WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:1020: The name tf.assign is deprecated. Please use tf.compat.v1.assign instead.

Epoch 1/10
131/131 [==============================] - 1124s 9s/step - loss: 2.7811 - val_loss: 2.1586
Epoch 2/10
131/131 [==============================] - 1108s 8s/step - loss: 2.0265 - val_loss: 1.9074
####################
Temperature: 0.2
####################
i could the coud you cand the goon

i love the with the love you cand the could to cour

i whe whe with a could the she with tour the soud I cone the love and the love the soud

####################
Temperature: 0.5
####################
i coud you me with a loood the cous I will the the the shen you coul to tour

wan I cand to me love the lowe I'll the criee

when don't the with you cone

####################
Temperature: 1.0
####################
dacl I'ly homendo gove thete aneaybine can by choum it yis coud

oh fouy coont trus

she your fell the love githor eplirine

Epoch 3/10
131/131 [==============================] - 1105s 8s/step - loss: 1.7636 - val_loss: 1.6658
Epoch 4/10
131/131 [==============================] - 1104s 8s/step - loss: 1.5593 - val_loss: 1.5089
####################
Temperature: 0.2
####################
i don't go me me

i'm gonna know you know you know

i know you've got to be make

####################
Temperature: 0.5
####################
baby nah nah nah nah nah nah

i'm grad me

you're goet to show you know

####################
Temperature: 1.0
####################
and need guiltad arone

she should dane

hele sleep it's liel me

Epoch 5/10
131/131 [==============================] - 1103s 8s/step - loss: 1.4092 - val_loss: 1.3872
Epoch 6/10
131/131 [==============================] - 1103s 8s/step - loss: 1.2847 - val_loss: 1.3122
####################
Temperature: 0.2
####################
i want to be the life on the man

i want to be the monely want

i want to be the money the way

####################
Temperature: 0.5
####################
and I'll all together now

i know you'll be like me a will you

when I'm going to me

####################
Temperature: 1.0
####################
shece I go a carem on

so child ways don't apparty can so

mo the pain to but me nobody

Epoch 7/10
131/131 [==============================] - 1099s 8s/step - loss: 1.1877 - val_loss: 1.2614
Epoch 8/10
131/131 [==============================] - 1099s 8s/step - loss: 1.1100 - val_loss: 1.2192
####################
Temperature: 0.2
####################
i saw a shame what a shame mary jane baby jud

i saw a show and she would say so should

i want to be your man and she said

####################
Temperature: 0.5
####################
the calling and name and the party

i saw a walk and I sad a show

i need to see you and me mine

####################
Temperature: 1.0
####################
is well you see you

there will getting better is understand

but as belight up an mind after my my hand

Epoch 9/10
131/131 [==============================] - 1102s 8s/step - loss: 1.0410 - val_loss: 1.1814
Epoch 10/10
131/131 [==============================] - 1105s 8s/step - loss: 0.9878 - val_loss: 1.1679
####################
Temperature: 0.2
####################
it's going to lose that you see

and I will sing the word of the same

and I will send you and I'll be and I'll be there

####################
Temperature: 0.5
####################
can't you see you thangering love boy

with in the evening

i thought a should can't be long

####################
Temperature: 1.0
####################
oh oh ah ah oh

don't let me didn't let me down

look she's meeting home

print(textgen.model.summary())

Model: "model_2"
__________________________________________________________________________________________________
Layer (type)                    Output Shape         Param #     Connected to                     
==================================================================================================
input (InputLayer)              (None, 40)           0                                            
__________________________________________________________________________________________________
embedding (Embedding)           (None, 40, 100)      4100        input[0][0]                      
__________________________________________________________________________________________________
rnn_1 (Bidirectional)           (None, 40, 256)      234496      embedding[0][0]                  
__________________________________________________________________________________________________
rnn_2 (Bidirectional)           (None, 40, 256)      394240      rnn_1[0][0]                      
__________________________________________________________________________________________________
rnn_3 (Bidirectional)           (None, 40, 256)      394240      rnn_2[0][0]                      
__________________________________________________________________________________________________
rnn_4 (Bidirectional)           (None, 40, 256)      394240      rnn_3[0][0]                      
__________________________________________________________________________________________________
rnn_concat (Concatenate)        (None, 40, 1124)     0           embedding[0][0]                  
                                                                 rnn_1[0][0]                      
                                                                 rnn_2[0][0]                      
                                                                 rnn_3[0][0]                      
                                                                 rnn_4[0][0]                      
__________________________________________________________________________________________________
attention (AttentionWeightedAve (None, 1124)         1124        rnn_concat[0][0]                 
__________________________________________________________________________________________________
output (Dense)                  (None, 41)           46125       attention[0][0]                  
==================================================================================================
Total params: 1,468,565
Trainable params: 1,468,565
Non-trainable params: 0
__________________________________________________________________________________________________
None

weights_path = '{}_weights.hdf5'.format(model_name)

vocab_path = '500nds_12Lrs_100epchs_Model_vocab (1).json'

config_path = '500nds_12Lrs_100epchs_Model_config.json'

textgen = textgenrnn(weights_path=weights_path,
                       vocab_path=vocab_path,
                       config_path=config_path)

generated_characters = 300

text = textgen.generate_samples(300)

####################
Temperature: 0.2
####################
i want you

i want to be your man I want to be your man

i want you so bad

you know you know my name

i want you

i want to be your man I me mine

i want to be your man and kissing the same of the bight the sun

i can't be and me mine

i want you so bad

you know I nearly had a pain at the party

i want you so bad

and I will sing the time of the same

i want you so hold you

i said someone who's got to be your man

i want you so bad

i want you so bad

i want you so bad

i want to be your man I'm true

i want to be your man I'm lonely

i want you so hold me too

when I saw you should see you make me the mind

i want you so bad

i want you

it's all too much the number

i want you

i want you so hold you man

when I was you see you and me

i want you so bad

i want you so bad

i want to be your man I know what you know my name

i want you

i can't be and me alone

it's true another man

i want you so bad

i want to be your man I don't know that you see

i want to be your man I me mine

i want you

the word the word there's no need

i want you so hold you head you can do

i want to be your man I can't see

i want to be your man I me mine

i want you so hold you

i want to be your man I me mine

i want you so bad

i want you so bad

i want you so hold me too much oh

i want to be your man I me mine

i want you

i want you

i want you so bad

the word the word of the start

i want you so bad

i want you so hard that you see

i want you so bad

i want you so bad

i can't believe you

i want you so hold you man

i want you so hold me too

i want you so bad

i want to be your man I me mine

i want you so bad

i want you so many many jane had a pain at the party

i want you so bad

i want you so bad

when I saw you should see the end

i want you so bad

i want you so bad

i want to be your man I me mine

i want you so bad

i'm gonna cread you there met me so mind

i'm tried to make you mean to me

i think you with another day

when I was a shame what a shame mary jane had a pain at the party

you know I nearly had a pain at the party

i want you so bad

i want you so bad

i want to be your man I me mine

i want you so hard the word there with a lot of the band

i can't be long to be long listen play

when I was you see you see you mean to me

i want you so bad

i want you so hold me too

i want you so bad

and I want to be your man 

i want you so bad

i want you you should see you know

it's all too much you yeah

i want you so bad

i want you so bad

i want you

i want you so bad

i want you so bad

i want to be your man I me mine

i want you so bad

i want you so bad

i want you

i want you so hold me too

i want you so hard the changes the same

i want you so hold me too

i want you so hard the way be baby cry

i want you

i want you so hold me there

i think you should see you see

i want to be your man I me mine

i want you so hold you

i want you so hard that I was a show

i want you so bad

when I saw you should see you want to be your man

i want to be your man I me mine

i want you so bad

and the word the sun she was a shame

i want you so bad

i want you so bad

all the world there's no need you

i want you

i want you so bad

it's all too much oh yeah

the world that she said so

i want you so bad

i want you so bad

i want you so bad

i want you you see you so make me

i want to be your man I me mine

i want you so bad

i want you so bad

i'm gonna get to be there with you

i want you so bad

i want you so bad

i want you so bad

i want you so bad

i want to be your man I me mine

i want you

i want you so bad

i can't be you now the number

i want you so bad

i can't be and I feel and I'll love you

i want you so bad

i want you so bad

i want to be your man I'm really down

i want you so bad

i want you so bad

i want to be your man I'm really down

i want you so bad

i want you so hold you oh

i want you so bad

i want you so hold your mind

i want you so bad

and I will send the time at all the time

i want to be your man why

i want you so hold you

the word the word there's no need

i want to be your man I me mine

i want you so bad

i want you so hold me too much to me

i want you so hold me that I'm so selly down

i want you so bad

i want you so bad

the word the word she said a love me when you believe been to your man

i want you so bad

i want to be your man I can do

i want you so bad

i want to be your man I me mine

and I want to be your man be man

i want you so bad

i want you so hold me too

when I was you see you there metter sing

i want you so bad

i want to be your man I me mine

i want you so bad

i want you so see you to me

i want you you should see you see

i want you so hold you

i want you

i want you so bad

i want you so bad

i want you so bad

the world the word the sun

i want you so bad

i want you so bad

when I was you see you believe me be long

i can't be baby beep beep me the wall

i want you so bad

i want you so hold you

i'm down I'm really down

i want to be your man I me mine

i want you so hold me too

i want you so hold you should

and I want to be your man I'm really down

i want you so bad

i want you so bad

i want you so hold you

the word the word there's no need

when I say he hard and feeling here

i can't be no no no no no no no

i want you so bad

i can't be there with a was a rich oh

i want you so bad

come on the beginning armary there

i want you so hard that you see

the world will spread the sun

i want to be your man 

i want you so bad

i want you so bad

it's all too much you

i want to be your mother should

i want you so hard that you know

i want you

i want you so bad

i can't be and mine

i want you so bad

i want you so hold you

i want you so many many many jane had a pain at the party

i want you so hard me

i want you so bad

i want you so bad

and I will see the word with a lot

i want you so hard that I was a shame mary jane what a shame mary jane had a pain at the party

i want you so bad

i want you so cry

i want you so bad

i want you so bad

i want to be your man I me mine

and I will see the word there's no need

i want to be your man I'm true

i want you so bad

i want to be your man I me mine

i can't be long long long to be bad you belong

i want you so bad

and I will send the world that you see it a pain of mind

i want you so hold your mind

i want you you should mean to me

i want to be your man I me mine

i want you so bad

i want you so bad

i want you

i want you so bad

i want you so bad

i want you so bad

you know I nearly hand you have and never been a long that line

i want you so bad

i want you you should see you mean to me

it's all too much yeah

she said so how me and I'll go

i want you so bad

i want you so bad

i want you so bad

i want to be your man I can't stay

and I will send the world I'm the world

when I was you see you love me down

and I want to be your man I'm down

and I would see you to make you

and I can see your money more

i want you

i want you so hold your mind

i want you so bad

and I will say the word there's no need

i want you

she was a shame what a shame mary jane had a pain at the party

when I saw you see you know my name

i want you so bad

i want to be your man I want to be your man

i want you so bad

i want you so bad

i want you so bad

i want you so bad

i can't go that you once belonged

she was a way the word of the band

i want you so bad

i want you so hold you

i want you so bad

i want you so bad

the love that's all the time to say you see

i want you so bad

i want you so hold me too

i want you so bad

and I will send you to me

i want to be your man I can't do that

i want you so bad

she said so should can be me

i think you should see you see

i want you so bad

i want you so bad

i want you so bad

i want you so bad

i want you

i want you so bad

i want you

i want to be your man I me mine

i want you so hard that I was a pain to me

i want you so hold you

i want you so bad

and I will send the time at all the time

i don't know what you know my name

i want you so hold you

i want you so hard that head

i want you so bad

i want you so hard the word the same on the party

the word the sun she was be baby just can't be mine

i want you so bad

i want to be your man I me long to you

####################
Temperature: 0.5
####################
i want you

you can do that I saw you and that she said

you know you know my name

the bagin love will you go

i want to be your studant so much in the night

you can talk to much you

come on when I have comes and the children

and I need the end

i know for the night before

but I'll go the end

he no one little girl

we didn't need me the night before

when you go arout the only window you had to go

i want you you should see you be bad

it could see you know I nearly buy

you know what you see it a partaction

i'm gonna get that you can be me

i want to be your man to make it a pain at a song on the dark

the word I want to be your man

i saw you should see me the one to me

i never been a long to be there

i can't talk you away yeah yeah yeah

i said and now I need you

so cry many man she had had easy

when I do you can see that you see

and then I'm all you see

tell me on love loved to be your life

i said

you know I'm down for your mind

can't you hold me and I have's gonna be a party what you see her too

love her

i said she said sunshine

i'm be my long to be long listen better

beep me I'm down the said

i would do that you've been me

i could never got to see you with yeah

and I'll be your man 

we're gonna stays where do a way be wrong

mother sump and when you know

see a book and I'll love you

when I get to be your man I me mine

i want you

all the wind I want the road

i said some how what it the same

it's all too hold me and I'm helalize

i'm gonna do you so much yeah

i'm be long you no but you'll be mine

i'm right all you near you soul

hello hello

i want to be your man 

i want you so bad

i can't be good on my love that's belonged

i'll get you near me the same with a was a star

don't worry

when I saw the man oh the same

and I will want to be no oh

i want to be your man come

want to be where you see you need

don't say here me so sad her belong

she said so hold me there before to me

i want you to day

i want you so hard she's gonna be too much 

i don't know what you see me me

will you in the one on the one and in the sun

i want to be your man I got to be your man

i want to be your man I me mine

and you know you know I think you that you know

now the sun sun now there is way

tell me when I am the way of mind

but I'm gonna be sad you see

it's all too much that like of yeah

she takes a father should want

you know I'm not there words of you

make a way be wants to cold you

baby you're so making and still the way

and she world the sun I feel again

there are no one love is all you need

i'm gonna get that I say going to me

i want to be the polled you will right

i want you

there's no need me to the long

so on I saw was to see you can't be mine

if I want you so hard to me

it's all the lonely penet to me

look at the old with you

i want you

you know that I want to be your man

i said a shord what I know how may

i'm looking through you where you coust to see the love is a place

and I will love you you can talk to me

treat me so long that have head

come on the bottom love girl window

when I tell you you near you don't know my nind hey by should

i can't be there one on the place in the party

when I wake there's on his nowhere

come on the bottom things are show

and the play back words of things is the sky with and everything I will do you old you can send that something I say be with a lot

i want to hold your hand

we'll never got a ticket to be a paperback writer

i want you

she was a little girl

i think you what you know my name

love you near me the man and there's and I feel the world

the world of the sun

i want you you so many touch

i'm coming oo so hold me there

i'm gonna be an our out her them

i want you you

and I will see you you can't be bad you

she keeps something away

let it be let it be bad out to be your mone

i want you

she said where you should can sleep tonight

i'm down I'm so so

i'm gonna get to see you more

i want you so comport me so man

there broken't let me down

come on let's nice him time

and the weater has back the start

i said there's no one me what a plann

i'm gonna carry that you know

i can't go out you and me and I want to be your man 

to carry to show you have to can all me

all the lonely better girl

and nothing to see you party change at his pulias

i want to be your man to be head

i can't believe you

just come on baby true

cause I hey have gone

help me was it is a with you

the way where are of the billed in the same

she world say she's a lot

sexy not a little girl from my friends

and I sing a little girl

say the love is here the begins

me the man what a shame mary jane wanded there with you

i want you

the word the way will seem me

yeah yeah yeah yeah yeah yeah

here such a love that show what I near yeah

i want you worry

i want to be your man to show

i don't know why hey love me shere it cry

the way thing will go

it's only only skenter wearing with a long to be me

you know I want you

all my number there's have a good from the long out of the end

and I want to be the one what you lied you

and I will I can see you that you know you know my milly feeling

there's a got a fool on the party

the world is the sky weres and so his nowhere

when I can't hide nowhere and post to the way

we do it is it's real

i want to hold your mind

you know you know my name

it's so a love with her it's sleeping

and I lost you you know you want me

i want you

when I was it the sone in the same

she's the tearing all the world there are going to see

i think you hear such oh

i can't a chair that have across of mind

they way be won't be sad feeling

i want you so bad

i can't be and mine

all the word there's a lonely

now I need to me

i want you so bad

you know what you say see

i want you you can be your mother should

yeah yeah yeah yeah yeah

and I will send the only time that have can comping out the band

well you can be and me my loving of when I say goodbye oh oh oh oh

and she said wead too

when I was you start the one

i want to be your eyes can't you door

she outs and heart a really one that with you

it's been a long to carolle

i got to be the sman I'm mad me ware

i can't get it be time to me

that is way there's no need

and I love you

good day so hold me in the fool

and I get the same and still wings that you do

because I'm the old you have got to have me a show

the tange at all the world that you do

i want you to stay that you see

the calles you say that mean a just before that

i want you so see her too

now I'm so the door of things of mind

the have what do disamet

so man I need no no oh head on the end

but broken the sun and I think you

all together now

you can't be the feet and long to you

did I take you and the way and I'm gonna be long or gain on

i want you so once paperback

the world that I still you

she said hello changile girl like a parter

so the sun going for your door

i can't believe me the same long to be make her me

yes I'm so leave it all too much me

the sun people better here

now I need to see the only girl to me

and I loved you

that you see it and I'll try too

hey gonna be a word of long

well you near me the word is that there

everybody can say him tance and I don't know

i want you

i know what I just be long to be long

and the things I look the place

i know I want you

i got no be a show of your money so

i don't know that she can she say hello

when you say you say the end I'm so the dark

why will you still the love and free

in the morning more me mad

i don't know that it's going to make you

i can't be long long little girl

if you know my name ha no need that hello

it's so and on the time of time

i can't be mine

i think around for you

you don't be long nothing you know that you do

dear much a want to be your man before to me

i'm going to lose you

she does it ain't you have me too

i want to be your man I love you

i want to be your mother baby just can't be mine

there's nothing you know my name

but I know I nearly say hello

the very and so mind you know

that she do do you no should

when I can't so boy

i want you

when you know I don't know my name

i'm down I'm true to misey to mean a drink

i think you want me

and he said morning morning

i'll be an the word of too much oh yeah

that good and so

all the lonely time is the sky

the word she was a millow good

the words and what it is you do

all together now

i think what you do

i want to be your man me away

the world of mind the word that she's happy just her help her

we do it is so criend rings of love

i want to see you that I can be hand

i could love for your night and I'd sad you belong

i love you yeah yeah yeah yeah

and I will wear well the said

the word is wondering

i said she would do

and I can see your money touce of mind

we can be what I should gonna go ou

she the coustion down the end

i'm so home of soo much

when you don't worry

and I will see you love me do

i feel you need you and me you're mine

when I saw you can dead the could

i think you know my name

i can't come home to say

all together now

someday someone

and I loved you

and the word is a with on more to me

it's my home

don't be long with you on the number

the calles in a lot to kiss you

but you know what you know I heard before out of her back ago to me

i'm gonna be the beginning

i want to be your man I think you should

you may see your mother should

i want you

you say all the windo that you see you

you know I need you

and I'm trying to make you

and I will do it's only playents

all together now

he said so hard in the life of the start

you know that you know my name

what oh she would leave me in the light

you'll be my mind

for the baginning of the mystery ain't be silver right from the blistle help me

you let me down you to let me know

i'm gonna get to show that I'm round

here got a little happened along

i want you

make you there mean me the word

but the more will go

oh yeah yeah yeah

now time is here it won't be mine

it's a loser

the sun my walls here

i feel you need to me

if you can see that I want me

####################
Temperature: 1.0
####################
everyone plamis her eyes

here can he can aperivad

tell me away sad you can

love me love new me the mack

kissona without out someone

just near right I know how man

in you know my will

well I'm stone the tartce play

well you inter's in driverstile wind me

'cause I'm love a woad a skilam

the sun we're on a lover hole

well you burn of you sunty face out

i said it criend from shine

we're looks to much the beginning

she would daday new there with the night before and been the sun

she love you toof

christer writced time thationing

i only hour them its and longer

now no namr named everything you know

come on the bird of good her

but go on colourther knyings me tires came 

because you know where yeah till I close are on I be

i can't sergline

oncs her life sun you leave me and mine

ain't you know that I'm telling

with a sooo on anyops

take your head back 

a look at the phone and knaw you silly stops I'm borwen

where I fell the wrong to what you can no

but you know I will the only 

and now I feel a way

she's got to die

she world papper as she said

love of all the little gight

and it's true peop you know

in the welt my boywen

far had their is nowhere my window

with you

i know if I want hey wrong

and you'll trupted to me

here gone only feelon

i said around down you know she won't be long

where change I wright dose hear you

honey my mine

there going down

but you know I nearly died

they wetermed nor like you

don't hidw the time without guy

i want you you near yeah yeah yeah

and then bant tonight

oh when I saw you didn't know without you

then you're too much what I see a

hello looked you make 

it's only in everything I'll ckild 

can't you see now tire the apaparazine 

in the love everyone frirty

help you trive to me

wonders do it apppear dayis has go words and hurther 

and I found time the oughtle no come 

so let you need is toptee another disappenebt never giving on up the show

come on let't papperback man

looking and I'll never make you

remember she feel

when I get you like tomorrow me

beby cry beep me writer to a par

all at the old on my mind of the sun

i'm gonna get a fut home

good make it all you need

since's you're reding nowhere to make you and I love you

so know I singing wo man't wind and fine

all your neah you

befper child she's be my one

then I'm going take from my right me

imside in the end

it looking in her

good us shoulder I'm in silver

you you say you mather bast broke

always fine me thinking of you

oh yesterday to so hurmas

a baby than the other

see he take you much and todays more to play

the his he walkad gone

i don't let you know I won't all you need

you'd gonna change goo gloou

when I really his were

it's herees born to be out cool

come on uncuyfoother had a fretend

come on the blits are love me me

her came it's giving walks low

celloo hil in I love you

in a mil out halt look

lace it's coming on the carme

don't give my will be dance ach don't lonely

oh oh i h ane have evened belong

so see you told me that I'm senthiee

it only times at armout tomorrown

look at home

it's all the hatchere

but she knows prease whoa I want to be in 

everybody I'm going to show out it's all only girl that boy

come together pack boy

server all together tone of mysteand

so baby that my finger people are I'm long

you burner the sun song you

honey didn't see you

and nothing I sad it come love morning

maded that some love and I'm a roa

when I feel you dyes a chance that you go

sitting roarshing someone or more

have another sudt her as good on bad good love

home for the little child

and you just finting so the walrus

when I so I feel good more

may's a wes blue

yes it way what weagh I'm kissing and ip

help me that is when you go

man't you realize it's real

we'd have a pain to give is in a man summer

don't let me real

i need " bitdaged darling

i'd inthing yight time to make me I'll be there

i need to me

take you out your near sin'm on minders

told me why you know my naw

who she said sun here had a said

but is the eggingen appen when you're insted and love girl that I said

you better better better them turns

martmaliss appartion

don't worret home to

did ky were baby with I belong

c'mon gives you

baby in and do

when the love any loned's bill spillacper

i want you

oh oh ano you silver eyes

there's nothing some

i am a boy had goesboot bag

ah ah ah oh

love thate's no near like a love

and lears where you dranks meettion

i don't need now

all I gotta do day

waiting down to give every just

there day with as the egg insterance"

one the little girl by mine

come on his love is what you see the oce

in mistereasplicter bloked to bill

just natten goodbye oncy love with me

elach you've got to say

couldn't have anyold I'm heard in really stops in her

she does at mind I'm gone

she one is madh a dranger before to hide and me

she will go it be

you told me un without of yestalnzaling

i've amilan guy you'ted hand before

cry brring look at the evening one your flirtantion I've bean

to please me coming baby 

look at home

she's got nothing you can good

where you have me seem armand

paperback little child

is who it show home about me in more

i sad all weight

oh my tump every don't care the man

come on letting he's your bot our provy

get back me love day sexy have ahearts in the ruin

i know you want you know how man

oh ah no no

but you know my name hate's halmie

now about it here such yeah like a with owhere

that's agoing there without alone

could me

if you'll never do

came it on simpore away

hey break it won't you hear you mum

if there's anyo good read

oh my changing the whaces she dog deads 

the newser I said hey'd kad to give or your say hurry with just bag little girl

everybody sunshine

once bring of little should

you know you see her hands inving a millex

love me earned alone

now I've minemed seeting all it tight

where you treat me see it be

love and friend out

if you look wonate everything you belong

sed inside you it's riverhin'

so is yeah can't see a remelatly hearts for easy 

torking out

my love without that before the touce blue

yes you know I've missteen con't much 

mn the many with go

oh no doe what's a rumb me had bom

let it be

anyctead is a just again out you

have that it kissoni-borncoom gonna come on

and I'm hof love is love is you see run

ah now I needed your birthday

you know you no ooh man ooh moo

that's shave wile gonely one

back where you're be cromning

remember where you'd be but little girl

it tell me alone good a

before me poot better make

martha will be narmown

come on his way a cride

and she was someone

if I think when I was alone bring 

oh pleasons in the sky name

come reople asteantil alrights

i say seed you stand after the party

made is procters to wrong

"mouls befora

you'll be gonna wanger hord inside

it the egg on

i can't tell my smore but the gues

that's that evening now na leave

send you didn't nead me in the ga

she longer me ow

back bydible after my proviess in the sgound of loww

little better to you to my feet of more

see the things the wild floe pain his out

but but if baby

well you did it a fa-la bla-la held do

hold me along how her dead

good just not tell you down

write that is in freen't alway

spease in the light

i me tall you sit compoad

someone the sun 

do

that you know homeone no now

she becked me mad to me

christmas is so my would ret

you know you no

i know you i had aloight to real

three I'm down I'm roadiy

boy you're gonna-moom

say you know my name nah on

it's so things to get good

i can't heep you were do

but it's all the word porgines

man polyong in the stree

perase not kind the bibtwas

i dig no can kndeen' night you make me hold

i'm just a show

as I new she changed your dosaine

we baby cry

seing me more day's girl

all around hearter leave me man

say to heaven mean to the blinks a face

good nudibng so thing but time

oh you feeling the cold

so the stag

how dreamers lough to wrong

here someone we lose it love

all together nows

you say

was it is will down you

teddy's rolly fadling I'm twice good must to heee a said doing

please let chile sail love

a the mun and with in 

here magicat feel be a pain't bed

i want you 

it baby head

did I want up 

someday you don't need

and a right come our your caccoo

if please don't have some now

i got a sain and happy pie

you shout to much eon't real me so my

i sady looks the cill like you

cry

you know my name

if you lifetime plesed

can't you go our in the man I want to be in my had

to be I've sween would low to be

he i'm an telling the one in my win

i'm down I'm silver blings in my eyes

it's always to gonna go

it's been to with on sounly away

ristucion fastio of you

you'll be in the nunging must plans

somelises homebod that prease don't badak

we eave it is my window girl poart

and it's making bast singing like it looking that playing closed

it's true you to smile

i'm just come oor to me

i couldn't be night and I've been is blue

"werepartand she was born

when I wask when I get you one love

everywhere away

sa-caling words on scrooution

they heavy needs you and I'm going to chack to be the hilleges

i think you love me a comfoas away

tell me nowwhe could never babe get nob letter

ob look yight been here's nothing to holk this back girl

all you need man somehel's maken in the dinkes of you

eight till me that you know me

feel up out on the party