Climate_site/python_scripts/companies_function_own_details.py

##packages code

import streamlit as st
import pandas as pd
import re
import json
from math import radians, cos, sin, asin, sqrt
import numpy as np

from sentence_transformers import SentenceTransformer

model = SentenceTransformer('all-MiniLM-L6-v2')


path = 'Climate_site/python_scripts/'


from sentence_transformers import SentenceTransformer, util

@st.cache_resource
def model_nlp():
    model = SentenceTransformer('all-MiniLM-L6-v2')
    return model


@st.cache_data
def load_dic():
    f = open(path + "preqin_venturedealsdetails_encoded.json","r")
    dic_companies = json.load(f)
    return dic_companies
    
@st.cache_data
def load_data():
    url = path + "preqin_companies_IEA.tsv"
    table = pd.read_csv(url, delimiter = "\t" , index_col = 0)
    table = table.astype({'portfolio_company_id': 'str'})

    return table




table_companies = load_data()
dic_companies = load_dic()





def norm(vector):
    return sqrt(sum(x * x for x in vector))    

def cosine_similarity2(vec_a, vec_b):
        norm_a = norm(vec_a)
        norm_b = norm(vec_b)
        dot = sum(a * b for a, b in zip(vec_a, vec_b))
        return dot / (norm_a * norm_b)



def get_similar_company(field, description , size):

    model = model_nlp()
    
    companies = dic_companies[field]
    
    
    encoded_description = model.encode(description)
    scores = np.dot( np.array(list(companies.values())) , np.array(encoded_description))
    
    
    dic_scores = { str(k) : { "score" : v }  for k , v in zip(companies.keys(), scores )}
    
    res = pd.DataFrame(dic_scores).T.sort_values("score" , ascending = False).head(size)
    
    return res.merge(table_companies , left_index = True , right_on = "portfolio_company_id" , how = "left")