logistic_regression_model (1).py

import pandas as pd
import numpy as np
import re, os, time, csv, random, pymongo, math, html, string, html2text
from pymongo import MongoClient
import xml.etree.ElementTree as ET
import pymongo
import sklearn



#Currently we have a pandas DataFrame with a row for each article classification
#The fields assumed when writing this code are 'grade', 'text', 'title', 'date', etc.

def get_lm_words():
    global lm_words
    lm_words = pd.read_csv('../../pluribus_labs/earnings_call/data/dictionary_data/LoughranMcDonald_MasterDictionary_2014.csv')['Word']
    return lm_words    

def get_tf_vector(text):
    text = text.upper()
    text_split = text.split()
    tf_dict = {}
    for word in lm_words:
        tf_dict[word] = pd.Series(text_split).map(lambda x: word == x).sum()
    tf_vector = tf_dict.values()
    tf_keys = tf_dict.keys()
    return tf_vector

def get_grade_tf_df(df):
    text_col = df.text
    df['tf_vector'] = text_col.map(lambda x: get_tf_vector(x))
    return df[['grade', 'tf_vector']]


def df_to_csv(df, path):
    return df.to_csv(path)    


def logistic_regression_fit(grade_tf_df):
    grade = grade_tf_df.grade
    tf_vector = grade_tf_df.tf_vector
    LR_model = sklearn.linear_model.LogisticRegression()
    
    
def logistic_regression_predict(LR_fit, tf_matrix):
    #The tf matrix is a list of tf_vectors you want a grade for
    #This functions returns a classification (grade) for each input tf_vector
    LR_predict = LR_fit.predict(tf_vector)
    return LR_predict

def logistic_regression_summary(LR_fit):
    return LR_fit.summary()