TopicModelingTrustpilotReviews/modelization.py at master · brash6/TopicModelingTrustpilotReviews · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
import heapq
import numpy as np
import plotly.graph_objects as go
from gensim import corpora
from gensim.models.ldamulticore import LdaMulticore
from gensim.models import CoherenceModel

import constants as cst
import data_cleaning
import visualizations


# Defining a function to get coherence score
def my_coherence_vals(dictionary, corpus, texts, limit, start, step):
    coherence_values = []
    model_list = []
    for num_topics in range(start, limit, step):
        model = LdaMulticore(corpus=corpus, num_topics=num_topics, id2word=dictionary,
                             chunksize=2000, passes=50, per_word_topics=True)
        model_list.append(model)
        coherencemodel = CoherenceModel(model=model, texts=texts, dictionary=dictionary, coherence='c_v')
        coherence_values.append(coherencemodel.get_coherence())

    return model_list, coherence_values


# Function to return and visualize best model in term of coherence
def viz_best_coherence(id2word, corpus, texts, start=4, limit=9, step=2):
    model_list, coherence_values = my_coherence_vals(id2word, corpus,
                                                     texts, limit, start, step)
    colors = ['lightslategray', ] * (limit - start)
    index_of_best = coherence_values.index(max(coherence_values))
    best_indexes = heapq.nlargest(2, range(len(coherence_values)), key=coherence_values.__getitem__)
    if (abs(coherence_values[best_indexes[0]] - coherence_values[best_indexes[1]]) < 0.01) and (
            best_indexes[0] < best_indexes[1]):
        index_of_best = best_indexes[1]
    colors[index_of_best] = 'crimson'
    fig = go.Figure(data=[go.Bar(
        x=np.arange(start, limit, step),
        y=coherence_values,
        marker_color=colors,
    )])
    fig.update_layout(title_text='Model coherence with different number of topics', width=1000, height=500)
    fig.show()
    return model_list[index_of_best]


def create_dict_corpus(df):
    reviews = df['review']
    # Create Dictionary
    id2word = corpora.Dictionary(reviews)
    # Create Corpus: Term Document Frequency
    corpus = [id2word.doc2bow(review) for review in reviews]
    return reviews, id2word, corpus


def create_dicts(time_periods, df_pos_reviews, df_neg_reviews):
    """
    time_periods : List of sorted dates which includes start and end dates, for example ["2015", "2017", "02-2020", "12-2020"]
    """
    df_dicts = {}
    df_dicts['pos'] = {}
    df_dicts['neg'] = {}
    for i in range(len(time_periods)):
        if i < len(time_periods) - 1:
            # Positive reviews
            df_dicts['pos'][time_periods[i] + '_' + time_periods[i + 1]] = {}
            df_dicts['pos'][time_periods[i] + '_' + time_periods[i + 1]]['df'] = df_pos_reviews[
                (df_pos_reviews['date'] >= time_periods[i])
                & (df_pos_reviews['date'] < time_periods[i + 1])]
            reviews_pos, id2word_pos, corpus_pos = \
                create_dict_corpus(df_dicts['pos'][time_periods[i] + '_' + time_periods[i + 1]]['df'])
            df_dicts['pos'][time_periods[i] + '_' + time_periods[i + 1]]['reviews'] = reviews_pos
            df_dicts['pos'][time_periods[i] + '_' + time_periods[i + 1]]['id2word'] = id2word_pos
            df_dicts['pos'][time_periods[i] + '_' + time_periods[i + 1]]['corpus'] = corpus_pos

            # Negative Reviews
            df_dicts['neg'][time_periods[i] + '_' + time_periods[i + 1]] = {}
            df_dicts['neg'][time_periods[i] + '_' + time_periods[i + 1]]['df'] = df_neg_reviews[
                (df_neg_reviews['date'] >= time_periods[i])
                & (df_neg_reviews['date'] < time_periods[i + 1])]
            reviews_neg, id2word_neg, corpus_neg = \
                create_dict_corpus(df_dicts['neg'][time_periods[i] + '_' + time_periods[i + 1]]['df'])
            df_dicts['neg'][time_periods[i] + '_' + time_periods[i + 1]]['reviews'] = reviews_neg
            df_dicts['neg'][time_periods[i] + '_' + time_periods[i + 1]]['id2word'] = id2word_neg
            df_dicts['neg'][time_periods[i] + '_' + time_periods[i + 1]]['corpus'] = corpus_neg
    return df_dicts


def train_ldas(time_periods, df_pos_reviews, df_neg_reviews):
    df_dicts = create_dicts(time_periods, df_pos_reviews, df_neg_reviews)
    for i in range(len(time_periods)):
        if i < len(time_periods) - 1:
            # Positive reviews
            print("Training LDA for positive reviews of " + time_periods[i] + '_' + time_periods[i + 1] + " period")
            df_dicts['pos'][time_periods[i] + '_' + time_periods[i + 1]]['model'] = viz_best_coherence(
                df_dicts['pos'][time_periods[i] + '_' + time_periods[i + 1]]['id2word'],
                corpus=df_dicts['pos'][time_periods[i] + '_' + time_periods[i + 1]]['corpus'],
                texts=df_dicts['pos'][time_periods[i] + '_' + time_periods[i + 1]]['reviews'])

            # Negative reviews
            print("Training LDA for negative reviews of " + time_periods[i] + '_' + time_periods[i + 1] + " period")
            df_dicts['neg'][time_periods[i] + '_' + time_periods[i + 1]]['model'] = viz_best_coherence(
                df_dicts['neg'][time_periods[i] + '_' + time_periods[i + 1]]['id2word'],
                corpus=df_dicts['neg'][time_periods[i] + '_' + time_periods[i + 1]]['corpus'],
                texts=df_dicts['neg'][time_periods[i] + '_' + time_periods[i + 1]]['reviews'])
    return df_dicts


def get_models_corpus_reviewslists(dict_models_over_time, pos=True):
    models = []
    corpus = []
    id2words = []
    reviews_lists = []
    if pos:
        for key in dict_models_over_time['pos']:
            models.append(dict_models_over_time['pos'][key]['model'])
            corpus.append(dict_models_over_time['pos'][key]['corpus'])
            id2words.append(dict_models_over_time['pos'][key]['id2word'])
            reviews_lists.append(dict_models_over_time['pos'][key]['reviews'])
    else:
        for key in dict_models_over_time['neg']:
            models.append(dict_models_over_time['neg'][key]['model'])
            corpus.append(dict_models_over_time['neg'][key]['corpus'])
            id2words.append(dict_models_over_time['neg'][key]['id2word'])
            reviews_lists.append(dict_models_over_time['neg'][key]['reviews'])
    return models, corpus, id2words, reviews_lists