mohdabdin

"""
This script is used to automate the process of sending LinkedIn connection requests and messages to a list of people.
It uses Selenium to automate the process of logging in and sending connection requests and messages.

Author: Mohammad Abdin (https://www.linkedin.com/in/mohammad-abdin-779260147/)

Keep in mind this can break if LinkedIn changes their HTML structure.
"""

import pandas as pd

mohdabdin

import yahoo_fin.stock_info as si

df = pd.read_csv('stocks.csv')

final_dic = {}
temp = si.get_quote_table('aapl')
for key in temp:
    final_dic.update({key:[]})

for idx, row in df.iterrows():

mohdabdin

while True:
    try:
        # Get Stock Card
        card = driver.find_element_by_xpath('/html/body/section/div/div/div[2]/div['+str(count)+']')
       
        # Extract Title
        stock_string = card.find_element_by_xpath(".//h6/div/span[2]/strong")
        driver.execute_script("return arguments[0].scrollIntoView();", stock_string)
        stock = stock_string.get_attribute('innerHTML')
        stock_symbol = stock.split()[-1][1:-1]

mohdabdin

import pandas as pd
import time

from selenium import webdriver
import selenium.webdriver.common.keys as kl
from selenium.webdriver.common.action_chains import ActionChains

driver = webdriver.Chrome("C:/chromedriver.exe")
driver.set_window_size(1280, 1080)
actions = ActionChains(driver)

mohdabdin

import gym
import time
from stable_baselines3 import A2C

def train(env):
    env = gym.make('CartPole-v0')
    model = A2C('MlpPolicy', env, verbose=1)
    model.learn(total_timesteps=20000)
    return model
    

mohdabdin

from policy_network import Policy

agent = Policy(obs_dim=4, act_dim=2)

for param in agent.parameters():
  print(param.data.shape)

mohdabdin

for epoch in range(EPOCHS):
    evolved_agents = [] #new population of agents to be populated
    # Perform Elitism
    for i in range(int(POP_SIZE*ELITISM_RATE)):
        evolved_agents.append(population.iloc[i]['agents'])

    while len(evolved_agents)<POP_SIZE:
        # Perform Selection
        parent_a, parent_b = selection(population)

mohdabdin

if __name__=='__main__':
    #initialize population
    agents = []
    for _ in range(POP_SIZE):
        agent = Policy(obs_dim=NUM_STATES, act_dim=NUM_ACTIONS)
        agents.append(agent)
    
    fitness_list = evaluate_agents(env, agents)
    
    population = pd.DataFrame({'agents': agents,

mohdabdin

import numpy as np
import pandas as pd
import gym
from evolution_functions import evaluate_agents, crossover, mutate, selection
import torch.nn as nn
import torch as T

from policy_network import Policy

EPOCHS = 100

mohdabdin

# Select two parents from population based on fitness
def selection(population):
    # Define a probability distribution proportional to each agent's fitness/reward score
    fit_sum = sum(population['fitness'])
    prob_dist = population['fitness']/fit_sum

    # Choose random indices from distribution making sure they're not equal
    par_a_idx = 0
    par_b_idx = 0
    while par_a_idx==par_b_idx: