|
1
|
|
|
import pandas as pd |
|
2
|
|
|
from numpy import array |
|
3
|
|
|
from sklearn.decomposition import PCA as sklearnPCA |
|
4
|
|
|
from sklearn.preprocessing import StandardScaler |
|
5
|
|
|
from sklearn.preprocessing import LabelEncoder |
|
6
|
|
|
from sklearn.preprocessing import OneHotEncoder |
|
7
|
|
|
from tabpy.models.utils import setup_utils |
|
8
|
|
|
|
|
9
|
|
|
|
|
10
|
|
|
def PCA(component, _arg1, _arg2, *_argN): |
|
11
|
|
|
""" |
|
12
|
|
|
Principal Component Analysis is a technique that extracts the key |
|
13
|
|
|
distinct components from a high dimensional space whie attempting |
|
14
|
|
|
to capture as much of the variance as possible. For more information |
|
15
|
|
|
on the function and how to use it please refer to tabpy-tools.md |
|
16
|
|
|
""" |
|
17
|
|
|
cols = [_arg1, _arg2] + list(_argN) |
|
18
|
|
|
encodedCols = [] |
|
19
|
|
|
labelEncoder = LabelEncoder() |
|
20
|
|
|
oneHotEncoder = OneHotEncoder(categories="auto", sparse=False) |
|
21
|
|
|
|
|
22
|
|
|
for col in cols: |
|
23
|
|
|
if isinstance(col[0], (int, float)): |
|
24
|
|
|
encodedCols.append(col) |
|
25
|
|
|
elif type(col[0]) is bool: |
|
26
|
|
|
intCol = array(col) |
|
27
|
|
|
encodedCols.append(intCol.astype(int)) |
|
28
|
|
|
else: |
|
29
|
|
|
if len(set(col)) > 25: |
|
30
|
|
|
print( |
|
31
|
|
|
"ERROR: Non-numeric arguments cannot have more than " |
|
32
|
|
|
"25 unique values" |
|
33
|
|
|
) |
|
34
|
|
|
raise ValueError |
|
35
|
|
|
integerEncoded = labelEncoder.fit_transform(array(col)) |
|
36
|
|
|
integerEncoded = integerEncoded.reshape(len(col), 1) |
|
37
|
|
|
oneHotEncoded = oneHotEncoder.fit_transform(integerEncoded) |
|
38
|
|
|
transformedMatrix = oneHotEncoded.transpose() |
|
39
|
|
|
encodedCols += list(transformedMatrix) |
|
40
|
|
|
|
|
41
|
|
|
dataDict = {} |
|
42
|
|
|
for i in range(len(encodedCols)): |
|
43
|
|
|
dataDict[f"col{1 + i}"] = list(encodedCols[i]) |
|
44
|
|
|
|
|
45
|
|
|
if component <= 0 or component > len(dataDict): |
|
46
|
|
|
print("ERROR: Component specified must be >= 0 and " "<= number of arguments") |
|
47
|
|
|
raise ValueError |
|
48
|
|
|
|
|
49
|
|
|
df = pd.DataFrame(data=dataDict, dtype=float) |
|
50
|
|
|
scale = StandardScaler() |
|
51
|
|
|
scaledData = scale.fit_transform(df) |
|
52
|
|
|
|
|
53
|
|
|
pca = sklearnPCA() |
|
54
|
|
|
pcaComponents = pca.fit_transform(scaledData) |
|
55
|
|
|
|
|
56
|
|
|
return pcaComponents[:, component - 1].tolist() |
|
57
|
|
|
|
|
58
|
|
|
|
|
59
|
|
|
if __name__ == "__main__": |
|
60
|
|
|
setup_utils.deploy_model("PCA", PCA, "Returns the specified principal component") |
|
61
|
|
|
|
tableau /
TabPy
