locker98/AI-Stock-Predictor
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
main
AI-Stock-Predictor/Train Stock Model LSTM Neural Network.ipynb
locker98 069c026d15 added code
2024-10-07 21:24:55 -04:00

156 KiB
Raw Permalink Blame History

None <html lang="en"> <head> </head>
In [1]:
import numpy as np
import matplotlib.pyplot as plt
import pandas as pd
import pandas_ta as ta
import IPython

# Uncomment for Interactive Graphs
#%matplotlib widget

name = "GDX"
data = pd.read_csv("data/"+name + ".csv")
data.tail(5)
Out[1]:
Open High Low Close Adj Close Volume Date
4608 39.80 40.22 39.77 40.09 40.09 27456000 24-09-13
4609 40.00 40.23 39.58 39.89 39.89 12162500 24-09-16
4610 39.70 40.08 39.31 39.49 39.49 17906900 24-09-17
4611 39.76 40.99 39.02 39.06 39.06 41241400 24-09-18
4612 40.10 40.25 39.26 39.72 39.72 22277500 24-09-19
In [2]:
# Adding indicators
data['Diff_Close'] = data['Adj Close']-data.Open
data['RSI']=ta.rsi(data.Close, length=14)
data['EMAF']=ta.ema(data.Close, length=18)
data['EMAM']=ta.ema(data.Close, length=50)
data['EMAS']=ta.ema(data.Close, length=200)
data.ta.bbands(append=True)
data.ta.stoch(append=True)


data['Target1'] = data['Diff_Close'].shift(-1)
data['Target2'] = data['Diff_Close'].shift(-2)
data['Target3'] = data['Diff_Close'].shift(-3)
data['Target4'] = data['Diff_Close'].shift(-4)
data['Target5'] = data['Diff_Close'].shift(-5)
data['Target6'] = data['Diff_Close'].shift(-6)
data['Target7'] = data['Diff_Close'].shift(-7)
data['Target8'] = data['Diff_Close'].shift(-8)
data['Target9'] = data['Diff_Close'].shift(-9)


data.dropna(inplace=True)
data.reset_index(inplace = True)
pd.set_option('display.max_columns', None)

data.tail(2)
Out[2]:
index Open High Low Close Adj Close Volume Date Diff_Close RSI EMAF EMAM EMAS BBL_5_2.0 BBM_5_2.0 BBU_5_2.0 BBB_5_2.0 BBP_5_2.0 STOCHk_14_3_3 STOCHd_14_3_3 Target1 Target2 Target3 Target4 Target5 Target6 Target7 Target8 Target9
4403 4602 37.66 37.83 37.28 37.33 37.33 16260000 24-09-05 -0.33 47.133189 37.965111 37.071605 33.759355 36.218919 37.802 39.385081 8.375644 0.350924 15.729597 32.100843 -0.85 0.15 0.46 0.49 1.15 0.29 -0.11 -0.21 -0.70
4404 4603 37.17 37.43 36.22 36.32 36.32 19939200 24-09-06 -0.85 41.326931 37.791941 37.042130 33.784834 35.778392 37.290 38.801608 8.107312 0.179150 9.755741 18.352188 0.15 0.46 0.49 1.15 0.29 -0.11 -0.21 -0.70 -0.38
In [3]:
data.drop(['Volume', 'Close', 'Date'], axis=1, inplace=True)
data_set = data
data_set.head()
Out[3]:
index Open High Low Adj Close Diff_Close RSI EMAF EMAM EMAS BBL_5_2.0 BBM_5_2.0 BBU_5_2.0 BBB_5_2.0 BBP_5_2.0 STOCHk_14_3_3 STOCHd_14_3_3 Target1 Target2 Target3 Target4 Target5 Target6 Target7 Target8 Target9
0 199 38.70 38.70 37.97 32.99 -5.71 42.823849 39.130266 39.242898 38.151400 36.894752 37.894 38.893248 5.273910 0.568051 26.968411 24.878202 -4.50 -6.22 -4.61 -4.51 -5.38 -4.97 -5.43 -4.69 -5.66
1 200 38.02 38.73 37.93 33.52 -4.50 46.633750 39.079712 39.219647 38.156361 37.773181 38.240 38.706819 2.441522 0.939142 29.608675 27.927078 -6.22 -4.61 -4.51 -5.38 -4.97 -5.43 -4.69 -5.66 -5.36
2 201 38.44 38.75 37.06 32.22 -6.22 39.735192 38.876584 39.138484 38.146348 37.049243 38.054 39.058757 5.280689 0.050140 23.950967 26.842684 -4.61 -4.51 -5.38 -4.97 -5.43 -4.69 -5.66 -5.36 -5.46
3 202 37.01 37.48 36.46 32.40 -4.61 41.049977 38.716944 39.068740 38.138523 36.764084 37.908 39.051916 6.035223 0.260472 20.792079 24.783907 -4.51 -5.38 -4.97 -5.43 -4.69 -5.66 -5.36 -5.46 -5.66
4 203 37.68 38.44 37.68 33.17 -4.51 46.333706 38.666739 39.036240 38.139533 36.773123 37.886 38.998877 5.874874 0.659047 20.107004 21.616683 -5.38 -4.97 -5.43 -4.69 -5.66 -5.36 -5.46 -5.66 -5.34
In [4]:
from sklearn.preprocessing import MinMaxScaler
sc = MinMaxScaler(feature_range=(-1,1))

df_scaled = sc.fit_transform(data_set.to_numpy())
data_set_scaled_pd = pd.DataFrame(df_scaled, columns=data_set.columns.tolist())



data_set_scaled_pd['index'] = data_set['index']

data_set_scaled_pd.tail()
Out[4]:
index Open High Low Adj Close Diff_Close RSI EMAF EMAM EMAS BBL_5_2.0 BBM_5_2.0 BBU_5_2.0 BBB_5_2.0 BBP_5_2.0 STOCHk_14_3_3 STOCHd_14_3_3 Target1 Target2 Target3 Target4 Target5 Target6 Target7 Target8 Target9
4400 4599 -0.032629 -0.035886 -0.032090 0.134366 0.357534 0.135110 -0.011064 -0.023456 -0.152467 -0.007310 -0.014559 -0.050959 -0.884847 -0.358144 0.330323 0.488259 0.279452 0.383562 0.335616 0.264384 0.401370 0.443836 0.447945 0.538356 0.420548
4401 4600 -0.061550 -0.068073 -0.085821 0.077857 0.279452 -0.115170 -0.015294 -0.023108 -0.150777 -0.047258 -0.029421 -0.041646 -0.762153 -0.905362 0.020945 0.291154 0.383562 0.335616 0.264384 0.401370 0.443836 0.447945 0.538356 0.420548 0.365753
4402 4601 -0.103077 -0.095819 -0.092537 0.062788 0.383562 -0.172337 -0.020589 -0.023368 -0.149273 -0.073525 -0.046937 -0.050252 -0.716935 -0.709322 -0.415359 -0.022089 0.335616 0.264384 0.401370 0.443836 0.447945 0.538356 0.420548 0.365753 0.352055
4403 4602 -0.074527 -0.078431 -0.071642 0.080368 0.335616 -0.090165 -0.023565 -0.022924 -0.147587 -0.082629 -0.054747 -0.056542 -0.709344 -0.299248 -0.697161 -0.372758 0.264384 0.401370 0.443836 0.447945 0.538356 0.420548 0.365753 0.352055 0.284932
4404 4603 -0.092696 -0.093230 -0.111194 0.038091 0.264384 -0.247492 -0.030463 -0.024164 -0.146390 -0.099513 -0.074158 -0.077581 -0.718980 -0.643250 -0.818783 -0.659348 0.401370 0.443836 0.447945 0.538356 0.420548 0.365753 0.352055 0.284932 0.328767
In [5]:
X = []
backcandles = 20


data_set_scaled = data_set_scaled_pd.to_numpy()

print("Length of Data", data_set_scaled.shape[0])

features = [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16]
#features = [5]
feature_count = len(features)

it = 0
for j in features:
    X.append([])
    for i in range(backcandles, data_set_scaled.shape[0]):
        X[it].append(data_set_scaled[i-backcandles:i, j])
    it += 1






#move axis from 0 to position 2
X=np.moveaxis(X, [0], [2])

X = np.array(X)

yi = np.array(data_set_scaled[backcandles:, -9:])
y=yi

print("X Shape:", X.shape)
print("Y Shape:", y.shape)

Length of Data 4405
X Shape: (4385, 20, 16)
Y Shape: (4385, 9)
In [6]:
# split data into train test sets
splitlimit = int(len(X)*1)
print(splitlimit)
X_train, X_test = X[:splitlimit], X[splitlimit:]
y_train, y_test = y[:splitlimit], y[splitlimit:]
print(X_train.shape)
print(X_test.shape)
print(y_train.shape)
print(y_test.shape)

4385
(4385, 20, 16)
(0, 20, 16)
(4385, 9)
(0, 9)
In [7]:
from keras.models import Sequential
from keras.layers import LSTM
from keras.layers import Dropout
from keras.layers import Dense
from keras.layers import TimeDistributed

import tensorflow as tf
import keras
from keras import optimizers
from keras.callbacks import History
from keras.models import Model
from keras.layers import Dense, Dropout, LSTM, Input, Activation, concatenate
import numpy as np

from tensorflow.keras.models import Sequential
from tensorflow.keras.optimizers import Adam
from tensorflow.keras import layers

np.random.seed(10)
print(X_train.shape)

model = Sequential([layers.LSTM(150, input_shape=(backcandles, feature_count), activation='tanh'),
                    layers.Dense(32, activation='relu'),
                    layers.Dense(9, name='dense_layer', activation='tanh')])


model.compile(loss='mse', 
              optimizer=Adam(learning_rate=0.001),
              metrics=['mean_absolute_error'])

model.summary()

epochs=2400
history = model.fit(x=X_train, y=y_train, batch_size=15, epochs=epochs, shuffle=True, validation_split = 0.2, verbose=0)

2024-10-07 20:29:09.720660: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:485] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered
2024-10-07 20:29:09.737019: E external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:8454] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
2024-10-07 20:29:09.741871: E external/local_xla/xla/stream_executor/cuda/cuda_blas.cc:1452] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered
2024-10-07 20:29:09.754590: I tensorflow/core/platform/cpu_feature_guard.cc:210] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
To enable the following instructions: AVX2 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
2024-10-07 20:29:10.373903: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT

(4385, 20, 16)

WARNING: All log messages before absl::InitializeLog() is called are written to STDERR
I0000 00:00:1728347350.828378  332220 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1728347350.874002  332220 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1728347350.878920  332220 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1728347350.883919  332220 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1728347350.887319  332220 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1728347350.891213  332220 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1728347351.030925  332220 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1728347351.032497  332220 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
I0000 00:00:1728347351.033977  332220 cuda_executor.cc:1015] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2024-10-07 20:29:11.035404: I tensorflow/core/common_runtime/gpu/gpu_device.cc:2021] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 5325 MB memory:  -> device: 0, name: NVIDIA GeForce RTX 3050, pci bus id: 0000:2d:00.0, compute capability: 8.6
/home/brickman/miniconda3/envs/stock/lib/python3.10/site-packages/keras/src/layers/rnn/rnn.py:204: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.
  super().__init__(**kwargs)

Model: "sequential"

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ lstm (LSTM)                     │ (None, 150)            │       100,200 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense (Dense)                   │ (None, 32)             │         4,832 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_layer (Dense)             │ (None, 9)              │           297 │
└─────────────────────────────────┴────────────────────────┴───────────────┘

 Total params: 105,329 (411.44 KB)

 Trainable params: 105,329 (411.44 KB)

 Non-trainable params: 0 (0.00 B)

2024-10-07 20:29:12.885047: I external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:531] Loaded cuDNN version 8907
In [8]:
acc = history.history['mean_absolute_error']
val_acc = history.history['val_mean_absolute_error']

loss = history.history['loss']
val_loss = history.history['val_loss']

epochs_range = range(epochs)

plt.figure(figsize=(11, 5))
plt.subplot(1, 2, 1)
plt.plot(epochs_range, acc, label='Training Accuracy')
plt.plot(epochs_range, val_acc, label='Validation Accuracy')
plt.legend(loc='lower right')
plt.title('Training and Validation Accuracy')

plt.subplot(1, 2, 2)
plt.plot(epochs_range, loss, label='Training Loss')
plt.plot(epochs_range, val_loss, label='Validation Loss')
plt.legend(loc='upper right')
plt.title('Training and Validation Loss')
plt.ion()
Out[8]:
<contextlib.ExitStack at 0x7cb97803ba30>
No description has been provided for this image
In [9]:
y_pred = model.predict(X_train)
for i in range(10):
    print("Pridicted:", y_pred[i], "Real value:", y_train[i])
print()
print(f"{y_train[-1]}")

138/138 ━━━━━━━━━━━━━━━━━━━━ 0s 2ms/step
Pridicted: [-0.4036709  -0.46461192 -0.3590175  -0.28324017 -0.34823442 -0.43334588
 -0.42882568 -0.41436017 -0.403928  ] Real value: [-0.40136986 -0.46164384 -0.35342466 -0.27945205 -0.34931507 -0.43013699
 -0.42739726 -0.41369863 -0.39726027]
Pridicted: [-0.46557218 -0.35770795 -0.2813072  -0.3510212  -0.42975435 -0.43653348
 -0.40678245 -0.4031321  -0.41870612] Real value: [-0.46164384 -0.35342466 -0.27945205 -0.34931507 -0.43013699 -0.42739726
 -0.41369863 -0.39726027 -0.41780822]
Pridicted: [-0.35209465 -0.28051785 -0.34810475 -0.43110624 -0.43116555 -0.41247454
 -0.39174688 -0.42331287 -0.42454508] Real value: [-0.35342466 -0.27945205 -0.34931507 -0.43013699 -0.42739726 -0.41369863
 -0.39726027 -0.41780822 -0.42328767]
Pridicted: [-0.28172114 -0.34396863 -0.4297893  -0.4320941  -0.41340354 -0.39519808
 -0.41816622 -0.42309865 -0.3611399 ] Real value: [-0.27945205 -0.34931507 -0.43013699 -0.42739726 -0.41369863 -0.39726027
 -0.41780822 -0.42328767 -0.36027397]
Pridicted: [-0.34929985 -0.42562777 -0.433488   -0.41463438 -0.39713734 -0.42165747
 -0.42167172 -0.3633375  -0.41996527] Real value: [-0.34931507 -0.43013699 -0.42739726 -0.41369863 -0.39726027 -0.41780822
 -0.42328767 -0.36027397 -0.42054795]
Pridicted: [-0.42858496 -0.4267884  -0.40962404 -0.39073083 -0.42235476 -0.4238721
 -0.3564185  -0.42203316 -0.3270994 ] Real value: [-0.43013699 -0.42739726 -0.41369863 -0.39726027 -0.41780822 -0.42328767
 -0.36027397 -0.42054795 -0.32739726]
Pridicted: [-0.43434742 -0.40459284 -0.38885665 -0.4140891  -0.42647222 -0.36434662
 -0.4164863  -0.32749146 -0.4512253 ] Real value: [-0.42739726 -0.41369863 -0.39726027 -0.41780822 -0.42328767 -0.36027397
 -0.42054795 -0.32739726 -0.45205479]
Pridicted: [-0.41408408 -0.3860175  -0.4147621  -0.42086872 -0.36401656 -0.41890422
 -0.32657167 -0.45092446 -0.3550458 ] Real value: [-0.41369863 -0.39726027 -0.41780822 -0.42328767 -0.36027397 -0.42054795
 -0.32739726 -0.45205479 -0.35342466]
Pridicted: [-0.39207783 -0.4143812  -0.42082974 -0.36102867 -0.42305246 -0.32819805
 -0.44679096 -0.35819703 -0.18476132] Real value: [-0.39726027 -0.41780822 -0.42328767 -0.36027397 -0.42054795 -0.32739726
 -0.45205479 -0.35342466 -0.18767123]
Pridicted: [-0.41665655 -0.41635478 -0.36310774 -0.41913277 -0.33059776 -0.4508693
 -0.35423303 -0.18578593 -0.30750462] Real value: [-0.41780822 -0.42328767 -0.36027397 -0.42054795 -0.32739726 -0.45205479
 -0.35342466 -0.18767123 -0.31643836]

[0.40136986 0.44383562 0.44794521 0.53835616 0.42054795 0.36575342
 0.35205479 0.28493151 0.32876712]
In [10]:
plt.figure(figsize=(10,5))
plt.plot(y_train[-1], color = 'orange', label = 'real')
plt.plot(y_pred[-1], color = 'green', label = 'pred')
plt.legend()
plt.ion()
Out[10]:
<contextlib.ExitStack at 0x7cb970beb0d0>
No description has been provided for this image

Save Model

In [11]:
model.save("models/" + name + ".keras")
</html>