from ballet import Feature
from ballet.eng import SimpleFunctionTransformer
input = ["Overall Qual", "Overall Cond"]
def calc_qual(df):
return df["Overall Qual"] - df["Overall Cond"]
transformer = SimpleFunctionTransformer(calc_qual)
name = "Qual"
feature = Feature(input=input, transformer=transformer, name=name)
from ballet import Feature
from ballet.eng import NullFiller, SimpleFunctionTransformer
input = ["Total Bsmt SF", "1st Flr SF", "2nd Flr SF"]
transformer = [
SimpleFunctionTransformer(lambda df: df.sum(axis=1)),
NullFiller()
]
name = "Total Area"
feature = Feature(input=input, transformer=transformer, name=name)
from ballet import Feature
from ballet.eng import SimpleFunctionTransformer
input = 'BsmtFin SF 2'
def has_qual(ser):
return (ser > 0).astype(int)
transformer = SimpleFunctionTransformer(has_qual)
name = 'Basement type 2 indicator'
feature = Feature(input=input, transformer=transformer, name=name)
from ballet import Feature
from ballet.eng import GroupwiseTransformer, SimpleFunctionTransformer
from sklearn.impute import SimpleImputer
input = ['Lot Frontage', 'Neighborhood']
transformer = [
SimpleFunctionTransformer(lambda df: df.set_index('Neighborhood',
append=True)),
GroupwiseTransformer(
SimpleImputer(strategy='median'),
groupby_kwargs={'level': 'Neighborhood'},
handle_error='ignore',
),
SimpleImputer(strategy='median'),
]
name = 'Imputed lot frontage by median per neighborhood'
feature = Feature(input=input, transformer=transformer, name=name)
+
b"AAAGNvcHlfWHEFiFgGAAAAbl9qb2JzcQZOWAUAAABjb2VmX3EHY251bXB5LmNvcmUubXVsd"
+
b"GlhcnJheQpfcmVjb25zdHJ1Y3QKcQhjbnVtcHkKbmRhcnJheQpxCUsAhXEKQwFicQuHcQxS"
+
b"cQ0oSwFLBoVxDmNudW1weQpkdHlwZQpxD1gCAAAAZjhxEEsASwGHcRFScRIoSwNYAQAAADx"
+
b"xE05OTkr/////Sv////9LAHRxFGKJQzCs1DbrA2ATwJ+EvJ2cgXZAgOdkq43NAUBPL1Ngi7"
+
b"xiwDcFZw6sG1VAsa1vEjH/c0BxFXRxFmJYCQAAAF9yZXNpZHVlc3EXY251bXB5LmNvcmUub"
+
b"XVsdGlhcnJheQpzY2FsYXIKcRhoEkMI3Eon+jyJq0JxGYZxGlJxG1gFAAAAcmFua19xHEsGWAkAAABzaW5ndWxhcl9xHW"
+
b"gIaAlLAIVxHmgLh3EfUnEgKEsBSwaFcSFoEolDMB/pi7yTBxpByuth8u3RrEDOZGHPHa6qQK+4epK+NaJADgShEyZ3mkC"
+
b"rrRv+vjeVQHEidHEjYlgKAAAAaW50ZXJjZXB0X3EkaBhoEkMIC+mioaOz/kBxJYZxJlJxJ1gQAAAAX3NrbGVhcm5fdmVy"
+ b"c2lvbnEoWAYAAAAwLjIyLjFxKXViLg==")
def base64_model(df):
magic = pickle.loads(base64.b64decode(secret))
X = df[keys].values
X[np.isnan(X)] = 0.0
return magic.predict(X)
input = keys
transformer = [SimpleFunctionTransformer(base64_model)]
name = "Base64 Model"
feature = Feature(input=input, transformer=transformer, name=name)
from ballet import Feature
from ballet.eng import SimpleFunctionTransformer
input = ["Overall Qual", "Overall Cond"]
transformer = SimpleFunctionTransformer(
lambda df: df["Overall Qual"] - df["Overall Cond"]
)
name = "Adjusted Overall Quality"
feature = Feature(input=input, transformer=transformer, name=name)
from ballet import Feature from ballet.eng import SimpleFunctionTransformer input = 'Open Porch SF' transformer = SimpleFunctionTransformer(lambda ser: ser > 0) name = 'Has open porch' feature = Feature(input=input, transformer=transformer, name=name)
import pandas as pd
from ballet import Feature
from ballet.eng import SimpleFunctionTransformer
from sklearn.preprocessing import OneHotEncoder
def calc_porch_type(df):
# Porch features
total_porch_area = df.sum(axis=1)
porch_type = pd.Series("Unknown", index=df.index)
porch_type[(total_porch_area == df["Enclosed Porch"])
& (df["Enclosed Porch"] > 0)] = "Enclosed"
porch_type[(total_porch_area == df["3Ssn Porch"])
& (df["3Ssn Porch"] > 0)] = "3Ssn"
porch_type[(total_porch_area == df["Open Porch SF"])
& (df["Open Porch SF"] > 0)] = "Open"
return porch_type
input = ["Enclosed Porch", "3Ssn Porch", "Open Porch SF"]
transformer = [
SimpleFunctionTransformer(calc_porch_type),
OneHotEncoder(),
]
name = "Porch Type (Cleaned, One-Hot Encoded)"
feature = Feature(input=input, transformer=transformer, name=name)
import numpy as np from ballet import Feature from ballet.eng import SimpleFunctionTransformer input = 'Lot Area' transformer = SimpleFunctionTransformer(np.sqrt) name = 'Sqrt Lot Area' feature = Feature(input=input, transformer=transformer, name=name)
import numpy as np
from ballet import Feature
from ballet.eng import SimpleFunctionTransformer
input = ["Lot Area", "Lot Frontage"]
def fill_frontage(df):
mask = df["Lot Frontage"].isnull()
df["Lot Frontage"][mask] = np.sqrt(df["Lot Area"])[mask]
return df["Lot Frontage"]
transformer = SimpleFunctionTransformer(fill_frontage)
name = "Lot Frontage Fill"
feature = Feature(input=input, transformer=transformer, name=name)
import pandas as pd
from ballet import Feature
from ballet.eng import SimpleFunctionTransformer
input = ["Year Built", "Garage Yr Blt"]
def calc_age(df):
return df["Year Built"].where(pd.notnull, df["Garage Yr Blt"])
transformer = SimpleFunctionTransformer(calc_age)
name = "Year built fill"
feature = Feature(input=input, transformer=transformer, name=name)
from ballet import Feature from ballet.eng import SimpleFunctionTransformer input = ["Full Bath", "Half Bath"] transformer = SimpleFunctionTransformer(lambda df: df["Full Bath"] + df["Half Bath"]) name = "Total Bath" feature = Feature(input=input, transformer=transformer, name=name)
from ballet import Feature
from ballet.eng import NullFiller, SimpleFunctionTransformer
input = ["Yr Sold", "Year Remod/Add"]
transformer = [
SimpleFunctionTransformer(lambda df: df["Yr Sold"] - df["Year Remod/Add"]),
NullFiller(),
]
name = "House Age"
feature = Feature(input=input, transformer=transformer, name=name)
from ballet import Feature
from ballet.eng import NullFiller, SimpleFunctionTransformer
input = ["Yr Sold", "Year Remod/Add"]
def calc_age(df):
return df["Yr Sold"] - df["Year Remod/Add"]
transformer = [SimpleFunctionTransformer(calc_age), NullFiller()]
name = "Age"
feature = Feature(input=input, transformer=transformer, name=name)
from ballet import Feature
from ballet.eng import NamedFramer, SimpleFunctionTransformer
from sklearn.preprocessing import OneHotEncoder
# Zeros in 020-090 get cut off. This feature prepends them back.
def pad_zero(ser):
return ser.astype(str).str.pad(3, side='left', fillchar='0')
input = 'MS SubClass'
transformer = [
SimpleFunctionTransformer(pad_zero),
NamedFramer('MS SubClass'),
OneHotEncoder(),
]
name = 'Pad zeros in MS SubClass'
feature = Feature(input=input, transformer=transformer, name=name)
from ballet import Feature from ballet.eng import NullFiller, SimpleFunctionTransformer input = ["Total Bsmt SF", "1st Flr SF", "2nd Flr SF"] transformer = [SimpleFunctionTransformer(lambda df: df.sum(axis=1)), NullFiller()] name = "Total Area Calculation" feature = Feature(input=input, transformer=transformer, name=name)