def beam_search(root, k, game_length): # k is the beam size
current_q = [root] # we will iterate through this to create child
next_q = [] # we will que children nodes here
df = data.tail(game_length + 1).copy()
count = 0
result = []
for level in range(game_length):
stockObject = p.Stock('AAPL', df.iloc[level, 0])
# stockObject = p.Stock('AAPL', df.iloc[game_length,0]) --> this is funny bug
# print("level:",game_length)
# print("open price on given day:",df.iloc[level,0])
while current_q:
node = current_q.pop()
for act in ['b', 's', 'h']: # buy sell hold
shares = int(root.five_percent / float(df.iloc[level][0]))
if doable(act, stockObject, node, shares):
count += 1
# print("action and shares",act,shares)
child = perform_action(act, node.player, stockObject,
shares)
node.children.append(child)
next_q.append(child)
if level != game_length - 1:
current_q = take_beams(k, next_q, level)
if level == game_length - 1:
final_result = take_beams(1, next_q, level)
# TotalShares = sum(playerObj.portfolio.values())
TotalCash = final_result[0].player.cash + \
(final_result[0].player.portfolio['AAPL'] * df.iloc[game_length, 0])
print("Cash", final_result[0].player.cash)
print("Portfolio", final_result[0].player.portfolio)
print("Last day Open Price", df.iloc[game_length, 0])
print("Total Cash Value", TotalCash)
print("Total Nodes Processed", count)
if TotalCash > root.starting_cash:
print(
f"Wow! our AI made profit of:{TotalCash - root.starting_cash}\n"
)
elif TotalCash < root.starting_cash:
print(
f"Oh No! our AI lost :{root.starting_cash - TotalCash}\n")
else:
print(f"No profit/ loss\n")
# return take_beams(1, next_q, level) # in the end of the game we will take the best node
result.append((TotalCash - root.starting_cash))
result.append(count)
return result
next_q = []
def basic_AI(root, k, game_length): # k is the beam size
current_q = [root] # we will iterate through this to create child
next_q = [] # we will que children nodes here
df = data.tail(game_length+1).copy()
count = 0
for level in range(game_length):
stockObject = p.Stock('AAPL', df.iloc[level,0])
while current_q:
node = current_q.pop()
for act in ['b', 's', 'h']: # buy sell hold
shares = int(root.five_percent/float(df.iloc[level][0]))
if doable(act, stockObject, node, shares):
count+=1
# new_node_obj = Node(action = act, player = p.Player(name = 'bob', money = copy.copy(node.player.cash), portf = copy.copy(node.player.portfolio)))
child = perform_action(act, node.player, stockObject, shares)
node.children.append(child)
next_q.append(child)
if level != game_length-1:
current_q = random_take_beam(k, next_q, level)
if level == game_length-1:
final_result = random_take_beam(1, next_q, level)
# TotalShares = sum(playerObj.portfolio.values())
TotalCash = final_result[0].player.cash + \
(final_result[0].player.portfolio['AAPL'] * df.iloc[game_length, 0])
print("Cash", final_result[0].player.cash)
print("Portfolio", final_result[0].player.portfolio)
print("Last day Open Price", df.iloc[game_length, 0])
print("Total Cash Value", TotalCash)
print("Total No of nodes processed:",count)
if TotalCash > root.starting_cash:
print(f"Wow! baseline AI made profit of:{TotalCash - root.starting_cash}\n")
elif TotalCash < root.starting_cash:
print(f"Oh No! baseline AI lost :{root.starting_cash - TotalCash}\n")
else:
print(f"No profit/ loss\n")
return (TotalCash - root.starting_cash) # in the end of the game we will take one node randomly
next_q = []
f"Player your Shares : {playerObj.portfolio} your Cash: {playerObj.cash}\n"
)
print(
f"Day {turns+1} apple stock: Open-->{float(df.iloc[turns][0])} \n"
)
print(f"High-->{float(df.iloc[turns][1])} \n")
print(f"Low-->{float(df.iloc[turns][2])} \n")
print(f"Close-->{float(df.iloc[turns][3])} \n")
print(f"Volume-->{float(df.iloc[turns][4])} \n")
print("4 things you can do")
print("1. Buy(B/b) 2. Sell(S/s) 3. Hold(H/h) 4. Exit(E/e) \n")
choice = input("what you want to do?!!!!\n")
#str1 = 'Apple'+str(turns)
stockObject = p.Stock('Apple', df.iloc[turns, 0])
if choice == 'S' or choice == 's':
#stck = input("which stock you want to sell?")
n_shares_sold = int(
input("How many shares you want to sell--> "))
if n_shares_sold > playerObj.portfolio['Apple']:
print(
f"Sorry you dont own that many stocks, Try again!!! \n"
)
continue
playerObj.sell(stockObject, int(n_shares_sold), 'Apple')
elif choice == 'B' or choice == 'b':
n_shares_bought = int(
input("How many shares you want to buy--> "))
def beam_search(root, k, game_length): # k is the beam size
current_q = [root] # we will iterate through this to create child
next_q = [] # we will que children nodes here
df = data.tail(game_length + 1).copy()
df_nn = data_nn.tail(game_length + 1).copy()
# b, h, s = data_nn['b'], data_nn['h'], data_nn['s']
# hot = np.column_stack((b, h, s))
# row_nn = hot[-level - 1]
# nn_action = ''
count = 0
y_targs = []
for level in range(game_length):
stockObject = p.Stock('AAPL', df.iloc[level, 0])
while current_q:
node = current_q.pop()
for act in ['b', 's', 'h']: # buy sell hold
# try:
shares = int(root.five_percent / float(df.iloc[level][0]))
# except ValueError:
if doable(act, stockObject, node, shares):
count += 1
new_node_obj = Node()
child = perform_action(act, node.player, stockObject,
shares)
node.children.append(child)
next_q.append(child)
if level != game_length - 1:
current_q = take_beams(k, next_q, level)
print(current_q)
y_targs.append(current_q[0].action)
if level == game_length - 1:
final_result = take_beams(1, next_q, level)
y_targs.append(final_result[0].action)
print(final_result)
# TotalShares = sum(playerObj.portfolio.values())
TotalCash = final_result[0].player.cash + \
(final_result[0].player.portfolio['AAPL'] * df.iloc[game_length, 0])
print("Cash", final_result[0].player.cash)
print("Portfolio", final_result[0].player.portfolio)
print("Last day Open Price", df.iloc[game_length, 0])
print("Total Cash Value", TotalCash)
print("Total no of nodes Processed:", count)
if TotalCash > root.starting_cash:
print(
f" Wow! our AI made profit of:{TotalCash - root.starting_cash}\n"
)
elif TotalCash < root.starting_cash:
print(
f"Oh No! our AI lost :{root.starting_cash - TotalCash}\n")
else:
print(f"No profit/ loss\n")
# return take_beams(1, next_q, level) # in the end of the game we will take the best node
# return (TotalCash - root.starting_cash)
next_q = []
return y_targs