microsoftml.concat: Sammanfogar flera kolumner till en enda vektor

Usage

microsoftml.concat(cols: [dict, list], **kargs)

Description

Kombinerar flera kolumner till en enda vektorvärdeskolumn.

Detaljer

concat skapar en enda vektorvärd kolumn från flera kolumner. Det kan utföras på data innan en modell tränar. Konkateneringen kan avsevärt påskynda databehandlingen när antalet kolumner är så stort som hundratals till tusentals.

Arguments

kolumner

En karaktärsdikt eller lista med variabelnamn att transformera. Om dict, representerar nycklarna namnen på nya variabler som ska skapas. Observera att alla indatavariabler måste vara av samma typ. Det är möjligt att producera flera utdatakolumner med sammanfogningstransformen. I det här fallet behöver du använda en lista med vektorer för att definiera en en-till-en-mappning mellan in- och utgångsvariabler. Till exempel, för att sammanfoga kolumnerna InNameA och InNameB till kolumnen OutName1 och även kolumnerna InNameC och InNameD till kolumnen OutName2, använd dict: dict(OutName1 = [InNameA, InNameB], outName2 = [InNameC, InNameD])

kargs

Ytterligare argument skickas till beräkningsmotorn.

Retur

Ett objekt som definierar konkateneringstransformen.

Example

'''
Example on logistic regression and concat.
'''
import numpy
import pandas
import sklearn
from microsoftml import rx_logistic_regression, concat, rx_predict
from microsoftml.datasets.datasets import get_dataset

iris = get_dataset("iris")

if sklearn.__version__ < "0.18":
    from sklearn.cross_validation import train_test_split
else:
    from sklearn.model_selection import train_test_split

# We use iris dataset.
irisdf = iris.as_df()

# The training features.
features = ["Sepal_Length", "Sepal_Width", "Petal_Length", "Petal_Width"]

# The label.
label = "Label"

# microsoftml needs a single dataframe with features and label.
cols = features + [label]

# We split into train/test. y_train, y_test are not used.
data_train, data_test, y_train, y_test = train_test_split(irisdf[cols], irisdf[label])

# We train a logistic regression.
# A concat transform is added to group features in a single vector column.
multi_logit_out = rx_logistic_regression(
                        formula="Label ~ Features",
                        method="multiClass",
                        data=data_train,
                        ml_transforms=[concat(cols={'Features': features})])
                        
# We show the coefficients.
print(multi_logit_out.coef_)

# We predict.
prediction = rx_predict(multi_logit_out, data=data_test)

print(prediction.head())

Resultat:

Automatically adding a MinMax normalization transform, use 'norm=Warn' or 'norm=No' to turn this behavior off.
Beginning processing data.
Rows Read: 112, Read Time: 0, Transform Time: 0
Beginning processing data.
Beginning processing data.
Rows Read: 112, Read Time: 0, Transform Time: 0
Beginning processing data.
Beginning processing data.
Rows Read: 112, Read Time: 0.001, Transform Time: 0
Beginning processing data.
LBFGS multi-threading will attempt to load dataset into memory. In case of out-of-memory issues, turn off multi-threading by setting trainThreads to 1.
Beginning optimization
num vars: 15
improvement criterion: Mean Improvement
L1 regularization selected 9 of 15 weights.
Not training a calibrator because it is not needed.
Elapsed time: 00:00:00.2348578
Elapsed time: 00:00:00.0197433
OrderedDict([('0+(Bias)', 1.943994402885437), ('1+(Bias)', 0.6346845030784607), ('2+(Bias)', -2.57867693901062), ('0+Petal_Width', -2.7277402877807617), ('0+Petal_Length', -2.5394322872161865), ('0+Sepal_Width', 0.4810805320739746), ('1+Sepal_Width', -0.5790582299232483), ('2+Petal_Width', 2.547518491744995), ('2+Petal_Length', 1.6753791570663452)])
Beginning processing data.
Rows Read: 38, Read Time: 0, Transform Time: 0
Beginning processing data.
Elapsed time: 00:00:00.0662932
Finished writing 38 rows.
Writing completed.
    Score.0   Score.1   Score.2
0  0.320061  0.504115  0.175825
1  0.761624  0.216213  0.022163
2  0.754765  0.215548  0.029687
3  0.182810  0.517855  0.299335
4  0.018770  0.290014  0.691216

Relaterat innehåll

  • Last updated on