forked from fbadine/LSTNet
-
Notifications
You must be signed in to change notification settings - Fork 0
/
run_forecast.py
159 lines (140 loc) · 6.45 KB
/
run_forecast.py
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
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
import tensorflow as tf
import os
import pickle
import json
import tqdm
from tqdm import tqdm
from tensorflow.keras.models import model_from_json
from LSTNet.lstnet_util import GetArguments, LSTNetInit
from LSTNet.lstnet_model import PreSkipTrans, PostSkipTrans, PreARTrans, PostARTrans, LSTNetModel, ModelCompile
import subprocess
import yfinance as yf
import pandas as pd
class FetchStocks():
def __init__(self,
period="1000d",
tickers=['AAPL', 'GOOGL', 'AMZN', 'MSFT', 'JPM', 'V', 'JNJ', 'PG', 'XOM', 'T', 'BAC', 'WMT', 'INTC', 'PFE',
'VZ', 'KO', 'TSLA', 'MRK', 'DIS', 'UNH', 'HD', 'ADBE', 'CMCSA', 'PEP', 'CSCO', 'NVDA', 'NFLX',
'ABT', 'NKE', 'CVX', 'ACN', 'TMUS', 'BMY', 'LLY', 'TMO', 'IBM', 'MCD', 'ORCL', 'UPS', 'MDT', 'COST',
'PM', 'AVGO', 'SAP', 'HON', 'NEE', 'TXN', 'MO'],
save_files=True
):
self.csv_path="LSTNet\data\large_portfolio.csv"
self.txt_path="LSTNet\data\large_portfolio.txt"
self.tickers=tickers
self.period=period
self.downloaded_data = yf.download(self.tickers, period=self.period,group_by='ticker', auto_adjust=True)
open_prices = pd.DataFrame({ticker: self.downloaded_data[ticker]['Open'] for ticker in self.tickers})
self.df=open_prices
if save_files==True :
open_prices.to_csv("LSTNet\data\large_portfolio.csv")
np.savetxt("LSTNet\data\large_portfolio.txt", np.array(open_prices), delimiter=',')
class LSTNetModel(FetchStocks):
def __init__ (self,
data_path="data\large_portfolio.txt",
horizon=1,
save_name="large_portfolio",
window=7,
validpercent=0.40,
batchsize=16,
skip=7,
epochs=100,
cnn_kernel=6,
learning_rate=0.001,
dropout=0.2,
highway=7,
GRUUnits=100,
SkipGRUUnits=5,
):
super().__init__()
self.data_path=data_path
self.horizon=horizon
self.save_name=save_name
self.window=window
self.validpercent=validpercent
self.batchsize=batchsize
self.skip=skip
self.epochs=epochs
self.cnn_kernel=cnn_kernel
self.learning_rate=learning_rate
self.dropout=dropout
self.highway=highway
self.GRUUnits=GRUUnits
self.SkipGRUUnits=SkipGRUUnits
def train_model(self):
print("training...")
os.chdir("LSTNet")
command = f'python main.py --data={self.data_path} --horizon={self.horizon} --save="save/{self.save_name}_horizon{self.horizon}_window{self.window}_skip{self.skip}" --test --savehistory --logfilename="log/lstnet" --debuglevel=20 --predict="all" --plot --save-plot="save/{self.save_name}_horizon{self.horizon}_window{self.window}_skip{self.skip}_plots" --window={self.window} --validpercent={self.validpercent} --batchsize={self.batchsize} --skip={self.skip} --epochs={self.epochs} --CNNKernel={self.cnn_kernel} --lr={self.learning_rate} --dropout={self.dropout} --highway={self.highway} --GRUUnits={self.GRUUnits} --SkipGRUUnits={self.SkipGRUUnits}'
process = subprocess.Popen(command, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
output, error = process.communicate()
if process.returncode == 0:
print(output.decode())
else:
print("Error:", error.decode())
print("model saved")
print(f'Outputs can be found at: LSTNet/save/{self.save_name}_horizon{self.horizon}_window{self.window}_skip{self.skip}')
def get_trained_model(self):
custom_objects = {"PreSkipTrans": PreSkipTrans,
"PostSkipTrans": PostSkipTrans,
"PreARTrans": PreARTrans,
"PostARTrans": PostARTrans,
}
json_file = open(f'save\{self.save_name}_horizon{self.horizon}_window{self.window}_skip{self.skip}.json', 'r')
loaded_model_json = json_file.read()
json_file.close()
loaded_model = model_from_json(loaded_model_json, custom_objects=custom_objects)
loaded_model.load_weights(f'save\{self.save_name}_horizon{self.horizon}_window{self.window}_skip{self.skip}.h5')
model = loaded_model
return model
class LSTNetIteratedModel(LSTNetModel, FetchStocks):
def __init__(self,
forecast_steps,
n_series,
csv_path,
save_forecast_json=True
):
super().__init__()
self.forecast_steps=forecast_steps
self.timesteps=self.window
self.series=np.array(pd.read_csv(csv_path))[:,1:]
self.json=save_forecast_json
def get_forecast(self):
model=self.get_trained_model()
series=self.series
time_steps=self.window
forecast_steps=self.forecast_steps
last_batch=series[-time_steps:,:]
forecast=[]
for step in range(forecast_steps):
pred=model.predict(np.array(last_batch, dtype='float32').reshape(1,time_steps,series.shape[1]))
forecast.append(pred)
last_batch=np.append(last_batch[1:,:],pred, axis=0)
forecast=np.array(forecast)
return forecast
def plot_lstnet_forecast(self, series_index=24):
model=self.get_trained_model()
series=self.series
time_steps=self.window
forecast=self.get_forecast()
forecast_df=pd.DataFrame(forecast.reshape(self.forecast_steps,series.shape[1]))
forecast_df.columns=self.tickers
series_df=pd.DataFrame(series)
series_df.columns=self.tickers
cumulative_df=pd.concat([series_df,forecast_df], axis=0)
cumulative_df=pd.DataFrame(np.array(cumulative_df))
cumulative_df.iloc[-(100+len(forecast_df)):-len(forecast_df),series_index].plot(color='blue')
cumulative_df.iloc[-len(forecast_df):,series_index].plot(color='red')
output_df=cumulative_df.iloc[-55:,:]
if self.json==True:
json_data = output_df.to_json(orient='index')
with open('LargePortfolioLSTNet_forecast.json', 'w') as f:
f.write(json_data)
print('JSON data saved to', 'LargePortfolioLSTNet_forecast.json')
plt.legend()
title_ticker=self.tickers[series_index]
plt.title(title_ticker)
plt.show()