sundar0989 / xuniverse Goto Github PK

Is it possible to use continuous outcome variable?

Can someone please explain how a vote is assigned or not assigned for each of those techniques used?
Is there a threshold in the feature importance or coefficients where it will render a vote to be assigned to a feature?

Thank you

I am trying to perform feature selection using xverse's VotingSelector on Titanic data set, which is a binary classification problem, and the dataset also contains categorical variables whcih i have one-hot-encoded. I am repeatedly facing this error. I am using xverse version 1.0.5 and the Python version is 3.6. Kindly help.

Would be nice if you could specify n_jobs=-1 to f.ex. to the voting selector so that it wouldn't just run in single core.

I can work around this problem by passing the entire dataframe in... along w/ the y, but I'm not sure if this is expected behavior. X implies passing dataframe without y included.

UnboundLocalError: local variable 'bins_X_grouped' referenced before assignment

df = pd.DataFrame({'x1': list(range(100)),
'x2': list(range(100)),
'x3': list(range(100)),
'y': list(range(100))})

df['y'] = np.where(df['x1']>50,1,0)

from xverse.transformer import MonotonicBinning
clf = MonotonicBinning()

X = df[['x1','x2','x3']]
y = df[['y']]

if not isinstance(X, pd.DataFrame):
print("Not a dataframe")
else:
print("Is a DataFrame")

clf.fit(X, y)

UnboundLocalError Traceback (most recent call last)
in
17 print("Is a DataFrame")
18
---> 19 clf.fit(X, y)

~/opt/anaconda3/lib/python3.7/site-packages/xverse/transformer/_binning.py in fit(self, X, y)
120
121 #apply the monotonic train function on dataset
--> 122 fit_X.apply(lambda x: self.train(x, y), axis=0)
123 return self
124

~/opt/anaconda3/lib/python3.7/site-packages/pandas/core/frame.py in apply(self, func, axis, raw, result_type, args, **kwds)
7766 kwds=kwds,
7767 )
-> 7768 return op.get_result()
7769
7770 def applymap(self, func, na_action: Optional[str] = None) -> DataFrame:

~/opt/anaconda3/lib/python3.7/site-packages/pandas/core/apply.py in get_result(self)
183 return self.apply_raw()
184
--> 185 return self.apply_standard()
186
187 def apply_empty_result(self):

~/opt/anaconda3/lib/python3.7/site-packages/pandas/core/apply.py in apply_standard(self)
274
275 def apply_standard(self):
--> 276 results, res_index = self.apply_series_generator()
277
278 # wrap results

~/opt/anaconda3/lib/python3.7/site-packages/pandas/core/apply.py in apply_series_generator(self)
288 for i, v in enumerate(series_gen):
289 # ignore SettingWithCopy here in case the user mutates
--> 290 results[i] = self.f(v)
291 if isinstance(results[i], ABCSeries):
292 # If we have a view on v, we need to make a copy because

~/opt/anaconda3/lib/python3.7/site-packages/xverse/transformer/_binning.py in (x)
120
121 #apply the monotonic train function on dataset
--> 122 fit_X.apply(lambda x: self.train(x, y), axis=0)
123 return self
124

~/opt/anaconda3/lib/python3.7/site-packages/xverse/transformer/_binning.py in train(self, X, y)
172 We still want our code to produce bins.
173 """
--> 174 if len(bins_X_grouped) == 1:
175 bins = algos.quantile(X, np.linspace(0, 1, force_bins)) #creates a new binnning based on forced bins
176 if len(np.unique(bins)) == 2:

UnboundLocalError: local variable 'bins_X_grouped' referenced before assignment

Hi,

Custome binning is not working , kindly advise.

can supply a parameter to control each category rate of the train samples?
Because when the category rate is very few,the binning will be meaningless.

ValueError: The input data must be pandas dataframe. But the input provided is <class 'str'>

I tried to debug but have been unable to figure out the reasoning behind this. Here's some sample code that recreates it:

df = pd.DataFrame({'x1': list(range(100)),
'x2': list(range(100)),
'y': list(range(100))})

df['y'] = np.where(df['x1']>50,1,0)

from xverse.transformer import MonotonicBinning
clf = MonotonicBinning()

X = df[['x1','x2']]
y = df[['y']]

if not isinstance(X, pd.DataFrame):
print("Not a dataframe")
else:
print("Is a DataFrame")

clf.fit(X, y)

ValueError Traceback (most recent call last)
in
10 print("Is a DataFrame")
11
---> 12 clf.fit(X, y)

~/opt/anaconda3/lib/python3.7/site-packages/xverse/transformer/_binning.py in fit(self, X, y)
76
77 #check datatype of X
---> 78 self.check_datatype(X)
79
80 #The length of X and Y should be equal

~/opt/anaconda3/lib/python3.7/site-packages/xverse/transformer/_binning.py in check_datatype(self, X)
62
63 if not isinstance(X, pd.DataFrame):
---> 64 raise ValueError("The input data must be pandas dataframe. But the input provided is " + str(type(X)))
65 return self
66

ValueError: The input data must be pandas dataframe. But the input provided is <class 'str'>

i change
bins = algos.quantile(X, np.linspace(0, 1, force_bins)) #creates a new binnning based on forced bins
to
bins = X.quantile(np.linspace(0, 1, force_bins)).values # creates a new binnning based on forced bins

Hi Sundar,

After running following code :
from xverse.transformer import MonotonicBinning
clf = MonotonicBinning()
clf.fit(X, y)

I am getting AttributeError: module 'pandas.core.algorithms' has no attribute 'quantile' error. Any help will be appreciated.

Thanks

TypeError Traceback (most recent call last)
in
----> 1 clf.transform(X).head()

~/.conda/envs/most-reg/lib/python3.7/site-packages/xverse/transformer/_woe.py in transform(self, X, y)
308 Estimator has to be fitted to apply transformations.")
309
--> 310 outX[new_column_name] = tempX.replace(self.woe_bins[original_column_name])
311
312 #transformed dataframe

~/.local/lib/python3.7/site-packages/pandas/core/series.py in replace(self, to_replace, value, inplace, limit, regex, method)
4567 limit=limit,
4568 regex=regex,
-> 4569 method=method,
4570 )
4571

~/.local/lib/python3.7/site-packages/pandas/core/generic.py in replace(self, to_replace, value, inplace, limit, regex, method)
6490
6491 return self.replace(
-> 6492 to_replace, value, inplace=inplace, limit=limit, regex=regex
6493 )
6494 else:

~/.local/lib/python3.7/site-packages/pandas/core/series.py in replace(self, to_replace, value, inplace, limit, regex, method)
4567 limit=limit,
4568 regex=regex,
-> 4569 method=method,
4570 )
4571

~/.local/lib/python3.7/site-packages/pandas/core/generic.py in replace(self, to_replace, value, inplace, limit, regex, method)
6536 dest_list=value,
6537 inplace=inplace,
-> 6538 regex=regex,
6539 )
6540

~/.local/lib/python3.7/site-packages/pandas/core/internals/managers.py in replace_list(self, src_list, dest_list, inplace, regex)
612 mask = ~isna(values)
613
--> 614 masks = [comp(s, mask, regex) for s in src_list]
615
616 result_blocks = []

~/.local/lib/python3.7/site-packages/pandas/core/internals/managers.py in (.0)
612 mask = ~isna(values)
613
--> 614 masks = [comp(s, mask, regex) for s in src_list]
615
616 result_blocks = []

~/.local/lib/python3.7/site-packages/pandas/core/internals/managers.py in comp(s, mask, regex)
606
607 s = com.maybe_box_datetimelike(s)
--> 608 return _compare_or_regex_search(values, s, regex, mask)
609
610 # Calculate the mask once, prior to the call of comp

~/.local/lib/python3.7/site-packages/pandas/core/internals/managers.py in _compare_or_regex_search(a, b, regex, mask)
1966 result = tmp
1967
-> 1968 _check_comparison_types(result, a, b)
1969 return result
1970

~/.local/lib/python3.7/site-packages/pandas/core/internals/managers.py in _check_comparison_types(result, a, b)
1934
1935 raise TypeError(
-> 1936 f"Cannot compare types {repr(type_names[0])} and {repr(type_names[1])}"
1937 )
1938

TypeError: Cannot compare types 'ndarray(dtype=object)' and 'Interval'

pandas version - 1.1.0

how to choose binning borders and number of bins
how better / different this from
https://github.com/airysen/caimcaim

It's better to add random_state param to VotingSelector for ExtraTreesClassifier and RandomForestClassifier, to be able reproduce results.