-
Notifications
You must be signed in to change notification settings - Fork 0
/
preprocessor.py
240 lines (196 loc) · 8.69 KB
/
preprocessor.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
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
# -*- coding: utf-8 -*-
"""
Created on Fri Aug 4 20:44:22 2023.
@author: Panagiotis Doupidis
"""
import pandas as pd
import numpy as np
from typing import Tuple, Union
from sklearn.preprocessing import StandardScaler
from sklearn.utils.validation import check_is_fitted
from sklearn.base import BaseEstimator, TransformerMixin
import torch
import os
import warnings
import joblib
import os.path
from utils import create_sequences
class FinancialPreprocessor:
"""
A class for preprocessing financial data for use with a machine learning model.
Parameters
----------
data : pd.DataFrame
The financial data to preprocess, as a pandas DataFrame.
test_size : int
The size of the test set.
scaler : BaseEstimator, TransformerMixin, optional
The scikit-learn scaler to use for scaling the data. Defaults to None.
"""
def __init__(self, data: pd.DataFrame, test_size: int, window: int,
scaler: BaseEstimator = None):
if not isinstance(data, pd.DataFrame):
raise ValueError("The data must be a pandas DataFrame")
self.data = data
self.window = window
self.test_size = test_size
self.scaler = scaler if scaler else StandardScaler().set_output(
transform='pandas')
# Check that the provided scaler is a valid scikit-learn scaler
if not isinstance(self.scaler, (BaseEstimator, TransformerMixin)):
raise ValueError(
"The provided scaler must be a valid scikit-learn scaler")
def get_train_data(self) -> Tuple[torch.Tensor, torch.Tensor]:
"""
Get preprocessed training data as PyTorch tensors.
Returns
-------
Tuple[torch.Tensor, torch.Tensor]
A tuple containing the preprocessed training data as tensors.
"""
# Split data into training set
train_data = self.data[:-self.test_size]
# Scale the data
train_data_scaled = self.scaler.fit_transform(train_data)
x_train, y_train = create_sequences(train_data_scaled, self.window)
# Convert data to PyTorch tensors
x_train = torch.from_numpy(x_train).float()
y_train = torch.from_numpy(y_train).float()
return x_train, y_train
def get_train_val_data(self,
val_fraction: float = 0.1) -> Tuple[torch.Tensor,
torch.Tensor,
torch.Tensor,
torch.Tensor]:
"""
Get preprocessed training and validation data as PyTorch tensors.
Parameters
----------
val_fraction : float
The size of the validation set as a fraction of the training set.
Returns
-------
Tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor]
A tuple containing the preprocessed training and validation data as
tensors.
"""
# Split data into training and test sets
train_data = self.data[:-self.test_size]
# Calculate the size of the validation set
self.val_size = int(len(train_data) * val_fraction)
if self.val_size < self.window:
raise ValueError(f'Validation size ({self.val_size}) cannot be < window ({self.window})\
. Consider increasing the fraction of data used for validation.')
# Split the training data into training and validation sets
train_data, val_data = train_data[:-
self.val_size], train_data[-self.val_size:]
# Scale the data
train_data_scaled = self.scaler.fit_transform(train_data)
val_data_scaled = self.scaler.transform(val_data)
x_train, y_train = create_sequences(train_data_scaled, self.window)
x_val, y_val = create_sequences(val_data_scaled, self.window)
# Convert data to PyTorch tensors
x_train = torch.from_numpy(x_train).float()
y_train = torch.from_numpy(y_train).float()
x_val = torch.from_numpy(x_val).float()
y_val = torch.from_numpy(y_val).float()
return x_train, y_train, x_val, y_val
def get_test_data(self) -> Tuple[torch.Tensor, torch.Tensor]:
"""
Get preprocessed testing data as PyTorch tensors.
Returns
-------
Tuple[torch.Tensor, torch.Tensor]
A tuple containing the preprocessed testing data as tensors.
"""
# Split data into testing set
test_data = self.data[-self.test_size:]
# Scale the data
self.test_data_scaled = self.scaler.transform(test_data)
x_test, y_test = create_sequences(self.test_data_scaled, self.window)
# Convert data to PyTorch tensors
x_test = torch.from_numpy(x_test).float()
y_test = torch.from_numpy(y_test).float()
return x_test, y_test
@staticmethod
def get_single(data: Union[pd.DataFrame, np.ndarray],
window_size: int,
scaler_path: Union[str, os.PathLike]) -> torch.Tensor:
"""
Convert input data to a float torch tensor.
Args
----
data(Union[pd.DataFrame, np.ndarray]): Input data, either a pandas
DataFrame or a 2D numpy array.
window_size(int): The number of rows that the input data must have.
scaler_path(Union[str, os.PathLike]): The path to a saved joblib
object containing a fitted scaler.
Returns
-------
torch.Tensor: The resulting float torch tensor.
Raises
------
AssertionError: If the input data is not a 2D numpy array or if it
does not have the specified number of rows.
ValueError: If the loaded scaler is not a valid estimator.
"""
if isinstance(data, pd.DataFrame):
data = data.to_numpy()
elif isinstance(data, np.ndarray):
assert data.ndim == 2, "Input numpy array must have 2 dimensions"
assert data.shape[0] == window_size, f"Input data must have {window_size} rows"
scaler = FinancialPreprocessor.load_scaler(scaler_path)
# Suppress any UserWarning raised when transforming the data
with warnings.catch_warnings():
warnings.simplefilter("ignore", UserWarning)
data = scaler.transform(data).to_numpy()
data = np.expand_dims(data[..., :-1], 0)
return torch.from_numpy(data).float()
@staticmethod
def load_scaler(scaler_path: str) -> Union[BaseEstimator,
TransformerMixin]:
"""
Load a fitted scaler from the specified file path.
Parameters
----------
scaler_path : str
The file path to load the fitted scaler from.
Returns
-------
scaler : BaseEstimator or TransformerMixin
The loaded fitted scaler.
Raises
------
FileNotFoundError
If the specified file path does not exist.
ValueError
If the loaded scaler is not a valid scikit-learn estimator/scaler.
"""
try:
with open(scaler_path, 'rb') as f:
scaler = joblib.load(f)
except FileNotFoundError:
raise FileNotFoundError('Path to scaler does not exist.')
# Check if the loaded scaler is a valid scikit-learn estimator/scaler
if not isinstance(scaler, (BaseEstimator, TransformerMixin)):
raise ValueError(
"The loaded scaler must be a valid scikit-learn estimator")
return scaler
def save_scaler(self, filename: str):
"""
Save a fitted instance of a scikit-learn StandardScaler to a file.
Args
----
filename(str): The name of the file to save the scaler to.
Raises
------
NotFittedError: If the scaler is not fitted.
"""
check_is_fitted(self.scaler)
# Check if the provided filename contains an extension
root, ext = os.path.splitext(filename)
# If an extension is present, replace it with the typical joblib ext.
filename = root + '.joblib' if not ext or ext != '.joblib' else filename
with open(filename, 'wb') as f:
joblib.dump(self.scaler, f)
print(f"The fitted scaler was saved to: {filename}")