matchillinois/backend.py at master · mnarayan1/matchillinois · 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
128
129
130
131
132
133
134
135
136
import pandas as pd
import numpy as np
import nltk
from nltk.corpus import stopwords
from sklearn.feature_extraction.text import CountVectorizer
from gensim import corpora, models
import gensim
from bertopic import BERTopic
import warnings
import csv

# NLP processing setup
nltk.download('stopwords')
stop_words = set(stopwords.words('english'))

def remove_stopwords(text):
    words = text.split()
    return " ".join(word for word in words if word.lower() not in stop_words)

def process_teams(input_csv_path, output_csv_path):
    data = pd.read_csv(input_csv_path)
    list_csv_files = []
    likert_columns = ["Web Dev", "ML", "Mobile Dev", "Game Dev", "Data Analytics"]
    weights = {1: 0.5**2, 2: 0.4**2, 3: 0.05**2, 4: 0.03**2, 5: 0.02**2}
    group_assignments = {}

    for idx, row in data.iterrows():
        topic_probabilities = [weights[val] for val in row[likert_columns]]
        total = sum(topic_probabilities)
        topic_probabilities = [p / total for p in topic_probabilities]
        user_id = row['User']
        assigned_topic_index = np.random.choice(len(topic_probabilities), p=topic_probabilities)
        assigned_topic = likert_columns[assigned_topic_index]
        group_assignments[user_id] = assigned_topic

    for topic in likert_columns:
        topic_data = [(user_id, data.loc[data['User'] == user_id, 'Project Description'].values[0])
                      for user_id, assigned_topic in group_assignments.items() if assigned_topic == topic]
        group_csv = f"{topic}_group.csv"
        pd.DataFrame(topic_data, columns=['User', 'Project Description']).to_csv(group_csv, index=False)
        list_csv_files.append(group_csv)


    def nlp_processing(file_name):
        data = pd.read_csv(file_name)
        data['Project Description'] = data['Project Description'].apply(remove_stopwords)
        texts = data['Project Description'].tolist()

        # Initialize LDATopic column to a default value
        data['LDATopic'] = -1

        # Fit BERTopic model
        bertopic_model = BERTopic(verbose=True, embedding_model='paraphrase-MiniLM-L3-v2', min_topic_size=5)
        bertopic_topics, _ = bertopic_model.fit_transform(texts)
        data['BERTopic'] = bertopic_topics

        # filter out outliers marked with -1 by BERTopic
        outlier_data = data[data['BERTopic'] == -1]
        outlier_texts = outlier_data['Project Description'].tolist()
        if not outlier_texts:
            # sort the data by the topic
            sorted_data = data.sort_values(by='BERTopic')
            sorted_data.to_csv(f"NLP_{file_name}", index=False)

        else:
            # preprocess for LDA
            vectorizer = CountVectorizer(stop_words='english')
            X = vectorizer.fit_transform(outlier_texts)
            corpus = gensim.matutils.Sparse2Corpus(X, documents_columns=False)
            dictionary = corpora.Dictionary.from_corpus(corpus, id2word=dict((id, word) for word, id in vectorizer.vocabulary_.items()))

            # fit LDA model on the outliers
            lda_model = gensim.models.LdaModel(corpus=corpus, id2word=dictionary, num_topics=5, passes=10)
            outlier_data['LDATopic'] = [max(lda_model[corpus[i]], key=lambda x: x[1])[0] for i in range(len(outlier_texts))]

            # combine the LDA topic assignments for outliers with the og dataset
            data.loc[data['BERTopic'] == -1, 'LDATopic'] = outlier_data['LDATopic']

            # sort the data first by BERTopic (excluding -1), then by LDATopic for outliers
            non_outlier_data = data[data['BERTopic'] != -1].sort_values(by='BERTopic')
            outlier_data_sorted = outlier_data.sort_values(by='LDATopic')
            sorted_data = pd.concat([non_outlier_data, outlier_data_sorted])


            sorted_data.to_csv(f"NLP_{file_name}", index=False)
        return f"NLP_{file_name}"

    nlp_csv_files = [nlp_processing(file) for file in list_csv_files]

    def create_teams(grouped_users):
        teams = []
        small_groups = []

        for group in grouped_users:
            while len(group) > 6:
                teams.append(group[:6])
                group = group[6:]
            if len(group) >= 3:
                teams.append(group)
            else:
                small_groups.extend(group)

        for member in small_groups:
            added_to_team = False
            for team in teams:
                if len(team) < 6:
                    team.append(member)
                    added_to_team = True
                    break
            if not added_to_team:
                teams.append([member])

        return teams

    final_data = []
    with open(output_csv_path, "w", newline='') as outfile:
        csv_writer = csv.writer(outfile)

        csv_writer.writerow(["Group", "Team Number", "User ID", "Project Description"])

        for file in nlp_csv_files:
            group_name = file.split('_')[1]

            data = pd.read_csv(file)
            grouped_lists = data.groupby('BERTopic').apply(lambda x: x['User'].tolist()).tolist()
            final_teams = create_teams(grouped_lists)

            for i, team in enumerate(final_teams):
                for user in team:
                    user_data = data[data['User'] == user].iloc[0]
                    team_data = [group_name, i + 1, user, user_data['Project Description']]
                    csv_writer.writerow(team_data)

                csv_writer.writerow([])

            csv_writer.writerow([])