I'm uploading my dataset, and I'm copying my dataset, but an error is appearing.
import numpy as np
import pandas as pd
import mathplotlib.pyplot as plt
house_data=pd.read_csv("/home/houseprice.csv")
#we evaluate the price of a house for those cases where the information is missing, for each variable
def analyse_na_value(df, var):
df - df.copy()
# we indicate as a variable as 1 where the observation is missing
# we indicate as 0 where the observation has a real value
df[var] = np.where(df[var].isnull(), 1 , 0)
#print(df[var].isnull())
# we calculate the mean saleprice where the information is missing or present
df.groupby(var)['SalePrice'].median().plot.bar()
plt.title(var)
plt.show()
for var in vars_with_na:
analyse_na_value(house_data, var)
error,when I comment this code line, I don't get an error
df - df.copy()
TypeError Traceback (most recent call last)
~/anaconda3/lib/python3.8/site-packages/pandas/core/ops/array_ops.py in na_arithmetic_op(left, right, op, is_cmp)
142 try:
--> 143 result = expressions.evaluate(op, left, right)
144 except TypeError:
~/anaconda3/lib/python3.8/site-packages/pandas/core/computation/expressions.py in evaluate(op, a, b, use_numexpr)
232 if use_numexpr:
--> 233 return _evaluate(op, op_str, a, b) # type: ignore
234 return _evaluate_standard(op, op_str, a, b)
~/anaconda3/lib/python3.8/site-packages/pandas/core/computation/expressions.py in _evaluate_numexpr(op, op_str, a, b)
118 if result is None:
--> 119 result = _evaluate_standard(op, op_str, a, b)
120
~/anaconda3/lib/python3.8/site-packages/pandas/core/computation/expressions.py in _evaluate_standard(op, op_str, a, b)
67 with np.errstate(all="ignore"):
---> 68 return op(a, b)
69
TypeError: unsupported operand type(s) for -: 'str' and 'str'
During handling of the above exception, another exception occurred:
TypeError Traceback (most recent call last)
<ipython-input-31-25d58bc46c86> in <module>
15
16 for var in vars_with_na:
---> 17 analyse_na_value(house_data, var)
<ipython-input-31-25d58bc46c86> in analyse_na_value(df, var)
1 #we evaluate the price of a house for those cases where the information is missing, for each variable
2 def analyse_na_value(df, var):
----> 3 df - df.copy()
4
5 # we indicate as a variable as 1 where the observation is missing
~/anaconda3/lib/python3.8/site-packages/pandas/core/ops/__init__.py in f(self, other, axis, level, fill_value)
649 if isinstance(other, ABCDataFrame):
650 # Another DataFrame
--> 651 new_data = self._combine_frame(other, na_op, fill_value)
652
653 elif isinstance(other, ABCSeries):
~/anaconda3/lib/python3.8/site-packages/pandas/core/frame.py in _combine_frame(self, other, func, fill_value)
5864 return func(left, right)
5865
-> 5866 new_data = ops.dispatch_to_series(self, other, _arith_op)
5867 return new_data
5868
~/anaconda3/lib/python3.8/site-packages/pandas/core/ops/__init__.py in dispatch_to_series(left, right, func, axis)
273 # _frame_arith_method_with_reindex
274
--> 275 bm = left._mgr.operate_blockwise(right._mgr, array_op)
276 return type(left)(bm)
277
~/anaconda3/lib/python3.8/site-packages/pandas/core/internals/managers.py in operate_blockwise(self, other, array_op)
362 Apply array_op blockwise with another (aligned) BlockManager.
363 """
--> 364 return operate_blockwise(self, other, array_op)
365
366 def apply(self: T, f, align_keys=None, **kwargs) -> T:
~/anaconda3/lib/python3.8/site-packages/pandas/core/internals/ops.py in operate_blockwise(left, right, array_op)
36 lvals, rvals = _get_same_shape_values(blk, rblk, left_ea, right_ea)
37
---> 38 res_values = array_op(lvals, rvals)
39 if left_ea and not right_ea and hasattr(res_values, "reshape"):
40 res_values = res_values.reshape(1, -1)
~/anaconda3/lib/python3.8/site-packages/pandas/core/ops/array_ops.py in arithmetic_op(left, right, op)
188 else:
189 with np.errstate(all="ignore"):
--> 190 res_values = na_arithmetic_op(lvalues, rvalues, op)
191
192 return res_values
~/anaconda3/lib/python3.8/site-packages/pandas/core/ops/array_ops.py in na_arithmetic_op(left, right, op, is_cmp)
148 # will handle complex numbers incorrectly, see GH#32047
149 raise
--> 150 result = masked_arith_op(left, right, op)
151
152 if is_cmp and (is_scalar(result) or result is NotImplemented):
~/anaconda3/lib/python3.8/site-packages/pandas/core/ops/array_ops.py in masked_arith_op(x, y, op)
90 if mask.any():
91 with np.errstate(all="ignore"):
---> 92 result[mask] = op(xrav[mask], yrav[mask])
93
94 else:
TypeError: unsupported operand type(s) for -: 'str' and 'str'
1
As far to what I know the copy() function works with python3,
but in pandas,
and python3 does it work I don't know.
How can I get rid of this error without commenting that code line?
I think you are supposed to do df = df.copy(). I would recommend changing the variable though. Here is an official Pandas documentation on this function. What you are doing is subtracting the data frame from itself...
Related
I'm trying this code for featuretools:
features, feature_names = ft.dfs(entityset = es, target_entity = 'demo',
agg_primitives = ['count', 'max', 'time_since_first', 'median', 'time_since_last', 'avg_time_between',
'sum', 'mean'],
trans_primitives = ['is_weekend', 'year', 'week', 'divide_by_feature', 'percentile'])
But I had this error
TypeError Traceback (most recent call last)
<ipython-input-17-89e925ff895d> in <module>
3 agg_primitives = ['count', 'max', 'time_since_first', 'median', 'time_since_last', 'avg_time_between',
4 'sum', 'mean'],
----> 5 trans_primitives = ['is_weekend', 'year', 'week', 'divide_by_feature', 'percentile'])
~/.local/lib/python3.6/site-packages/featuretools/utils/entry_point.py in function_wrapper(*args, **kwargs)
44 ep.on_error(error=e,
45 runtime=runtime)
---> 46 raise e
47
48 # send return value
~/.local/lib/python3.6/site-packages/featuretools/utils/entry_point.py in function_wrapper(*args, **kwargs)
36 # call function
37 start = time.time()
---> 38 return_value = func(*args, **kwargs)
39 runtime = time.time() - start
40 except Exception as e:
~/.local/lib/python3.6/site-packages/featuretools/synthesis/dfs.py in dfs(entities, relationships, entityset, target_entity, cutoff_time, instance_ids, agg_primitives, trans_primitives, groupby_trans_primitives, allowed_paths, max_depth, ignore_entities, ignore_variables, seed_features, drop_contains, drop_exact, where_primitives, max_features, cutoff_time_in_index, save_progress, features_only, training_window, approximate, chunk_size, n_jobs, dask_kwargs, verbose, return_variable_types)
226 n_jobs=n_jobs,
227 dask_kwargs=dask_kwargs,
--> 228 verbose=verbose)
229 return feature_matrix, features
~/.local/lib/python3.6/site-packages/featuretools/computational_backends/calculate_feature_matrix.py in calculate_feature_matrix(features, entityset, cutoff_time, instance_ids, entities, relationships, cutoff_time_in_index, training_window, approximate, save_progress, verbose, chunk_size, n_jobs, dask_kwargs)
265 cutoff_df_time_var=cutoff_df_time_var,
266 target_time=target_time,
--> 267 pass_columns=pass_columns)
268
269 feature_matrix = pd.concat(feature_matrix)
~/.local/lib/python3.6/site-packages/featuretools/computational_backends/calculate_feature_matrix.py in linear_calculate_chunks(chunks, feature_set, approximate, training_window, verbose, save_progress, entityset, no_unapproximated_aggs, cutoff_df_time_var, target_time, pass_columns)
496 no_unapproximated_aggs,
497 cutoff_df_time_var,
--> 498 target_time, pass_columns)
499 feature_matrix.append(_feature_matrix)
500 # Do a manual garbage collection in case objects from calculate_chunk
~/.local/lib/python3.6/site-packages/featuretools/computational_backends/calculate_feature_matrix.py in calculate_chunk(chunk, feature_set, entityset, approximate, training_window, verbose, save_progress, no_unapproximated_aggs, cutoff_df_time_var, target_time, pass_columns)
341 ids,
342 precalculated_features=precalculated_features_trie,
--> 343 training_window=window)
344
345 id_name = _feature_matrix.index.name
~/.local/lib/python3.6/site-packages/featuretools/computational_backends/utils.py in wrapped(*args, **kwargs)
35 def wrapped(*args, **kwargs):
36 if save_progress is None:
---> 37 r = method(*args, **kwargs)
38 else:
39 time = args[0].to_pydatetime()
~/.local/lib/python3.6/site-packages/featuretools/computational_backends/calculate_feature_matrix.py in calc_results(time_last, ids, precalculated_features, training_window)
316 ignored=all_approx_feature_set)
317
--> 318 matrix = calculator.run(ids)
319 return matrix
320
~/.local/lib/python3.6/site-packages/featuretools/computational_backends/feature_set_calculator.py in run(self, instance_ids)
100 precalculated_trie=self.precalculated_features,
101 filter_variable=target_entity.index,
--> 102 filter_values=instance_ids)
103
104 # The dataframe for the target entity should be stored at the root of
~/.local/lib/python3.6/site-packages/featuretools/computational_backends/feature_set_calculator.py in _calculate_features_for_entity(self, entity_id, feature_trie, df_trie, full_entity_df_trie, precalculated_trie, filter_variable, filter_values, parent_data)
187 columns=columns,
188 time_last=self.time_last,
--> 189 training_window=self.training_window)
190
191 # Step 2: Add variables to the dataframe linking it to all ancestors.
~/.local/lib/python3.6/site-packages/featuretools/entityset/entity.py in query_by_values(self, instance_vals, variable_id, columns, time_last, training_window)
271
272 if columns is not None:
--> 273 df = df[columns]
274
275 return df
~/.local/lib/python3.6/site-packages/pandas/core/frame.py in __getitem__(self, key)
2686 return self._getitem_multilevel(key)
2687 else:
-> 2688 return self._getitem_column(key)
2689
2690 def _getitem_column(self, key):
~/.local/lib/python3.6/site-packages/pandas/core/frame.py in _getitem_column(self, key)
2693 # get column
2694 if self.columns.is_unique:
-> 2695 return self._get_item_cache(key)
2696
2697 # duplicate columns & possible reduce dimensionality
~/.local/lib/python3.6/site-packages/pandas/core/generic.py in _get_item_cache(self, item)
2485 """Return the cached item, item represents a label indexer."""
2486 cache = self._item_cache
-> 2487 res = cache.get(item)
2488 if res is None:
2489 values = self._data.get(item)
TypeError: unhashable type: 'set'
I also tried the simplest code for deep feature synthesis (dfs) as shown below, but it still encountered the same error
features, feature_names = ft.dfs(entityset = es, target_entity = 'demo')
I'm not really sure why I encountered this error, any help or recommendations on how to go about from here is deeply appreciated.
Thanks in advance for your help!
I found a solution, my current version had bugs in it that was fixed by the FeatureTools team. Just run pip install directly from master,
pip install --upgrade https://github.com/featuretools/featuretools/zipball/master
This fixed and has been released in Featuretools 0.9.1. If you upgrade to the latest version of Featuretools, it will go away.
I have data like the sample below, which has 4 continuous columns [x0 to x3] and a binary column y. y has two values 1.0 and 0.0. I’m trying to check for correlation between the binary column y and one of the continuous columns x0, using the CatConCor function below, but I’m getting the error message below. The function creates a linear regression model and calcs the p value for the residuals with and without the categorical variable. If anyone can please point out the issue or how to fix it, it would be very much appreciated.
Data:
x_r x0 x1 x2 x3 y
0 0 0.466726 0.030126 0.998330 0.892770 0.0
1 1 0.173168 0.525810 -0.079341 -0.112151 0.0
2 2 -0.854467 0.770712 0.929614 -0.224779 0.0
3 3 -0.370574 0.568183 -0.928269 0.843253 0.0
4 4 -0.659431 -0.948491 -0.091534 0.706157 0.0
Code:
import numpy as np
import pandas as pd
from time import time
import scipy.stats as stats
from IPython.display import display # Allows the use of display() for DataFrames
# Pretty display for notebooks
%matplotlib inline
###########################################
# Suppress matplotlib user warnings
# Necessary for newer version of matplotlib
import warnings
warnings.filterwarnings("ignore", category = UserWarning, module = "matplotlib")
#
# Display inline matplotlib plots with IPython
from IPython import get_ipython
get_ipython().run_line_magic('matplotlib', 'inline')
###########################################
import matplotlib.pyplot as plt
import matplotlib.cm as cm
# correlation between categorical variable and continuous variable
def CatConCor(df,catVar,conVar):
import statsmodels.api as sm
from statsmodels.formula.api import ols
# subsetting data for one categorical column and one continuous column
data2=df.copy()[[catVar,conVar]]
data2[catVar]=data2[catVar].astype('category')
mod = ols(conVar+'~'+catVar,
data=data2).fit()
aov_table = sm.stats.anova_lm(mod, typ=2)
if aov_table['PR(>F)'][0] < 0.05:
print('Correlated p='+str(aov_table['PR(>F)'][0]))
else:
print('Uncorrelated p='+str(aov_table['PR(>F)'][0]))
# checking for correlation between categorical and continuous variables
CatConCor(df=train_df,catVar='y',conVar='x0')
Error:
---------------------------------------------------------------------------
AssertionError Traceback (most recent call last)
<ipython-input-6-80f83b8c8e14> in <module>()
1 # checking for correlation between categorical and continuous variables
2
----> 3 CatConCor(df=train_df,catVar='y',conVar='x0')
<ipython-input-2-35404ba1d697> in CatConCor(df, catVar, conVar)
103
104 mod = ols(conVar+'~'+catVar,
--> 105 data=data2).fit()
106
107 aov_table = sm.stats.anova_lm(mod, typ=2)
~/anaconda2/envs/py36/lib/python3.6/site-packages/statsmodels/base/model.py in from_formula(cls, formula, data, subset, drop_cols, *args, **kwargs)
153
154 tmp = handle_formula_data(data, None, formula, depth=eval_env,
--> 155 missing=missing)
156 ((endog, exog), missing_idx, design_info) = tmp
157
~/anaconda2/envs/py36/lib/python3.6/site-packages/statsmodels/formula/formulatools.py in handle_formula_data(Y, X, formula, depth, missing)
63 if data_util._is_using_pandas(Y, None):
64 result = dmatrices(formula, Y, depth, return_type='dataframe',
---> 65 NA_action=na_action)
66 else:
67 result = dmatrices(formula, Y, depth, return_type='dataframe',
~/anaconda2/envs/py36/lib/python3.6/site-packages/patsy/highlevel.py in dmatrices(formula_like, data, eval_env, NA_action, return_type)
308 eval_env = EvalEnvironment.capture(eval_env, reference=1)
309 (lhs, rhs) = _do_highlevel_design(formula_like, data, eval_env,
--> 310 NA_action, return_type)
311 if lhs.shape[1] == 0:
312 raise PatsyError("model is missing required outcome variables")
~/anaconda2/envs/py36/lib/python3.6/site-packages/patsy/highlevel.py in _do_highlevel_design(formula_like, data, eval_env, NA_action, return_type)
163 return iter([data])
164 design_infos = _try_incr_builders(formula_like, data_iter_maker, eval_env,
--> 165 NA_action)
166 if design_infos is not None:
167 return build_design_matrices(design_infos, data,
~/anaconda2/envs/py36/lib/python3.6/site-packages/patsy/highlevel.py in _try_incr_builders(formula_like, data_iter_maker, eval_env, NA_action)
60 "ascii-only, or else upgrade to Python 3.")
61 if isinstance(formula_like, str):
---> 62 formula_like = ModelDesc.from_formula(formula_like)
63 # fallthrough
64 if isinstance(formula_like, ModelDesc):
~/anaconda2/envs/py36/lib/python3.6/site-packages/patsy/desc.py in from_formula(cls, tree_or_string)
162 tree = tree_or_string
163 else:
--> 164 tree = parse_formula(tree_or_string)
165 value = Evaluator().eval(tree, require_evalexpr=False)
166 assert isinstance(value, cls)
~/anaconda2/envs/py36/lib/python3.6/site-packages/patsy/parse_formula.py in parse_formula(code, extra_operators)
146 tree = infix_parse(_tokenize_formula(code, operator_strings),
147 operators,
--> 148 _atomic_token_types)
149 if not isinstance(tree, ParseNode) or tree.type != "~":
150 tree = ParseNode("~", None, [tree], tree.origin)
~/anaconda2/envs/py36/lib/python3.6/site-packages/patsy/infix_parser.py in infix_parse(tokens, operators, atomic_types, trace)
208
209 want_noun = True
--> 210 for token in token_source:
211 if c.trace:
212 print("Reading next token (want_noun=%r)" % (want_noun,))
~/anaconda2/envs/py36/lib/python3.6/site-packages/patsy/parse_formula.py in _tokenize_formula(code, operator_strings)
92 else:
93 it.push_back((pytype, token_string, origin))
---> 94 yield _read_python_expr(it, end_tokens)
95
96 def test__tokenize_formula():
~/anaconda2/envs/py36/lib/python3.6/site-packages/patsy/parse_formula.py in _read_python_expr(it, end_tokens)
42 origins = []
43 bracket_level = 0
---> 44 for pytype, token_string, origin in it:
45 assert bracket_level >= 0
46 if bracket_level == 0 and token_string in end_tokens:
~/anaconda2/envs/py36/lib/python3.6/site-packages/patsy/util.py in next(self)
330 else:
331 # May raise StopIteration
--> 332 return six.advance_iterator(self._it)
333 __next__ = next
334
~/anaconda2/envs/py36/lib/python3.6/site-packages/patsy/tokens.py in python_tokenize(code)
33 break
34 origin = Origin(code, start, end)
---> 35 assert pytype not in (tokenize.NL, tokenize.NEWLINE)
36 if pytype == tokenize.ERRORTOKEN:
37 raise PatsyError("error tokenizing input "
AssertionError:
Upgrading patsy to 0.5.1 fixed the issue. I found the tip here:
https://github.com/statsmodels/statsmodels/issues/5343
I’m new to the cvxpy package. I’m trying to use it to work through an example from the following blog:
https://towardsdatascience.com/integer-programming-in-python-1cbdfa240df2
Where we’re trying to optimize the combination of marketing channels sent to a customer.
There’s been some recent changes to the cvxpy package and I’m getting the error below when I try to run the sum_entries step, (which has in the latest version been changed to cvxpy.sum)
I think the problem is coming from the dimensions of “selection” and “TRANSFORMER” being incompatible, but I’m not familiar enough with the cvxpy package to know. Any tips are greatly appreciated.
Code:
test_probs.shape
(200, 8)
Code:
# selection = cvxpy.Bool(*test_probs.shape) # syntax changed in latest version
selection = cvxpy.Variable(*test_probs.shape, boolean=True)
# constraints
# Constant matrix that counts how many of each
# material we sent to each customer
TRANSFORMER = np.array([[1,0,0],
[0,1,0],
[0,0,1],
[1,1,0],
[1,0,1],
[0,1,1],
[1,1,1],
[0,0,0]])
# can't send customer more promotion than there is supply
# note: sum_entries changed to sum in latest cvxpy version
supply_constraint = cvxpy.sum(selection * TRANSFORMER, axis=0) <= supply
Error:
---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
<ipython-input-47-f2ebf41a00af> in <module>()
18 # note: sum_entries changed to sum in latest cvxpy version
19
---> 20 supply_constraint = cvxpy.sum(selection * TRANSFORMER, axis=0) <= supply
21
22
~/anaconda2/envs/py36/lib/python3.6/site-packages/cvxpy/expressions/expression.py in cast_op(self, other)
47 """
48 other = self.cast_to_const(other)
---> 49 return binary_op(self, other)
50 return cast_op
51
~/anaconda2/envs/py36/lib/python3.6/site-packages/cvxpy/expressions/expression.py in __mul__(self, other)
385 return cvxtypes.multiply_expr()(self, other)
386 elif self.is_constant() or other.is_constant():
--> 387 return cvxtypes.mul_expr()(self, other)
388 else:
389 warnings.warn("Forming a nonconvex expression.")
~/anaconda2/envs/py36/lib/python3.6/site-packages/cvxpy/atoms/affine/binary_operators.py in __init__(self, lh_exp, rh_exp)
41
42 def __init__(self, lh_exp, rh_exp):
---> 43 super(BinaryOperator, self).__init__(lh_exp, rh_exp)
44
45 def name(self):
~/anaconda2/envs/py36/lib/python3.6/site-packages/cvxpy/atoms/atom.py in __init__(self, *args)
42 self.args = [Atom.cast_to_const(arg) for arg in args]
43 self.validate_arguments()
---> 44 self._shape = self.shape_from_args()
45 if len(self._shape) > 2:
46 raise ValueError("Atoms must be at most 2D.")
~/anaconda2/envs/py36/lib/python3.6/site-packages/cvxpy/atoms/affine/binary_operators.py in shape_from_args(self)
107 """Returns the (row, col) shape of the expression.
108 """
--> 109 return u.shape.mul_shapes(self.args[0].shape, self.args[1].shape)
110
111 def is_atom_convex(self):
~/anaconda2/envs/py36/lib/python3.6/site-packages/cvxpy/utilities/shape.py in mul_shapes(lh_shape, rh_shape)
140 lh_old = lh_shape
141 rh_old = rh_shape
--> 142 lh_shape, rh_shape, shape = mul_shapes_promote(lh_shape, rh_shape)
143 if lh_shape != lh_old:
144 shape = shape[1:]
~/anaconda2/envs/py36/lib/python3.6/site-packages/cvxpy/utilities/shape.py in mul_shapes_promote(lh_shape, rh_shape)
107 if lh_mat_shape[1] != rh_mat_shape[0]:
108 raise ValueError("Incompatible dimensions %s %s" % (
--> 109 lh_shape, rh_shape))
110 if lh_shape[:-2] != rh_shape[:-2]:
111 raise ValueError("Incompatible dimensions %s %s" % (
ValueError: Incompatible dimensions (1, 200) (8, 3)
Update:
I tried changing the selection shape as suggested in the comment below.
code:
selection = cvxpy.Variable(test_probs.shape, boolean=True)
and now I get the new error when I run the supply_constraint part of the code below.
code:
# constraints
# Constant matrix that counts how many of each
# material we sent to each customer
TRANSFORMER = np.array([[1,0,0],
[0,1,0],
[0,0,1],
[1,1,0],
[1,0,1],
[0,1,1],
[1,1,1],
[0,0,0]])
# can't send customer more promotion than there is supply
# note: sum_entries changed to sum in latest cvxpy version
supply_constraint = cvxpy.sum(selection * TRANSFORMER, axis=0) <= supply
Error:
---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
<ipython-input-10-6eb7a55ea896> in <module>()
18 # note: sum_entries changed to sum in latest cvxpy version
19
---> 20 supply_constraint = cvxpy.sum(selection * TRANSFORMER, axis=0) <= supply
21
22
~/anaconda2/envs/py36/lib/python3.6/site-packages/cvxpy/expressions/expression.py in cast_op(self, other)
47 """
48 other = self.cast_to_const(other)
---> 49 return binary_op(self, other)
50 return cast_op
51
~/anaconda2/envs/py36/lib/python3.6/site-packages/cvxpy/expressions/expression.py in __le__(self, other)
482 """NonPos : Creates an inequality constraint.
483 """
--> 484 return NonPos(self - other)
485
486 def __lt__(self, other):
~/anaconda2/envs/py36/lib/python3.6/site-packages/cvxpy/expressions/expression.py in cast_op(self, other)
47 """
48 other = self.cast_to_const(other)
---> 49 return binary_op(self, other)
50 return cast_op
51
~/anaconda2/envs/py36/lib/python3.6/site-packages/cvxpy/expressions/expression.py in __sub__(self, other)
370 """Expression : The difference of two expressions.
371 """
--> 372 return self + -other
373
374 #_cast_other
~/anaconda2/envs/py36/lib/python3.6/site-packages/cvxpy/expressions/expression.py in cast_op(self, other)
47 """
48 other = self.cast_to_const(other)
---> 49 return binary_op(self, other)
50 return cast_op
51
~/anaconda2/envs/py36/lib/python3.6/site-packages/cvxpy/expressions/expression.py in __add__(self, other)
358 """Expression : Sum two expressions.
359 """
--> 360 return cvxtypes.add_expr()([self, other])
361
362 #_cast_other
~/anaconda2/envs/py36/lib/python3.6/site-packages/cvxpy/atoms/affine/add_expr.py in __init__(self, arg_groups)
34 # For efficiency group args as sums.
35 self._arg_groups = arg_groups
---> 36 super(AddExpression, self).__init__(*arg_groups)
37 self.args = []
38 for group in arg_groups:
~/anaconda2/envs/py36/lib/python3.6/site-packages/cvxpy/atoms/atom.py in __init__(self, *args)
42 self.args = [Atom.cast_to_const(arg) for arg in args]
43 self.validate_arguments()
---> 44 self._shape = self.shape_from_args()
45 if len(self._shape) > 2:
46 raise ValueError("Atoms must be at most 2D.")
~/anaconda2/envs/py36/lib/python3.6/site-packages/cvxpy/atoms/affine/add_expr.py in shape_from_args(self)
42 """Returns the (row, col) shape of the expression.
43 """
---> 44 return u.shape.sum_shapes([arg.shape for arg in self.args])
45
46 def expand_args(self, expr):
~/anaconda2/envs/py36/lib/python3.6/site-packages/cvxpy/utilities/shape.py in sum_shapes(shapes)
50 raise ValueError(
51 "Cannot broadcast dimensions " +
---> 52 len(shapes)*" %s" % tuple(shapes))
53
54 longer = shape if len(shape) >= len(t) else t
ValueError: Cannot broadcast dimensions (3,) (1, 3)
Your issue is happening when you create the selection variable. You are unpacking the shape tuple into multiple arguments. The first argument to Variable should be a shape. So the correct construction is:
selection = cvxpy.Variable(test_probs.shape, boolean=True)
You can verify this is correct by inspecting the shape attribute:
selection.shape
Which should now give:
(200, 8)
I am using a while loop to calculate a cost function for memory reasons. When calculating the gradient, tensorflow will store Nm tensors where Nm is the number of iterations in my while loop (this cuases the same memory issues I had with the original energy functions). I do not want that as I don't have enough memory. So I want to register a new op along with a gradient function that both use a while loop. However I am having issues with using function.defun and a while loop. To simplify things, I have a small test example below:
import numpy as np
import tensorflow as tf
from tensorflow.python.framework import ops
from tensorflow.python.ops import array_ops
from tensorflow.python.ops import sparse_ops
from tensorflow.python.framework import function
def _run(tensor):
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
res = sess.run(tensor)
return res
#function.Defun(tf.float32,tf.float32,func_name ='tf_test_log')#,grad_func=tf_test_logGrad)
def tf_test_log(t_x,t_y):
#N = t_x.shape[0].value
condition = lambda i,m1: i<N
def body(index,x):
#return[(index+1),tf.concat([x, tf.expand_dims(tf.exp( tf.add( t_x[:,index],t_y[:,index]) ),1) ],1 ) ]
return[(index+1),tf.add(x, tf.exp( tf.add( t_x[:,0],t_y[:,0]) ) ) ]
i0 = tf.constant(0,dtype=tf.int32)
m0 = tf.zeros([N,1],dType)
ijk_0 = [i0,m0]
L,t_log_x = tf.while_loop(condition,body,ijk_0,
shape_invariants=[i0.get_shape(),
tf.TensorShape([N,None])]
)
return t_log_x
dType = tf.float32
N = np.int32(100)
t_N = tf.constant(N,dtype = tf.int32)
t_x = tf.constant(np.random.randn(N,N),dtype = dType)
t_y = tf.constant(np.random.randn(N,N),dtype = dType)
ys = _run(tf_test_log(t_x,t_y))
I then try to test the new op:
I get a Value error: The shape for while/Merge_1:0 is not an invariant for the loop. It enters the loop with shape (100, ?), but has shape after one iteration. Provide shape invariants using either the shape_invariants argument of tf.while_loop or set_shape() on the loop variables.
Note that calling
If i use a concatenate operation (instead of the add operation that gets returned by my while loop), I do not get any issues.
However, If I do not set N as a global variable (i.e. I do N = t_x.shape[0]) inside the body of the tf_test_log function, I get a Value error.
ValueError: Cannot convert a partially known TensorShape to a Tensor: (?, 1)
What is wrong with my code? Any help is greatly appreciated!
I am using python 3.5 on ubuntu 16.04 and tensorflow 1.4
full output:
ValueError Traceback (most recent call last)
~/Documents/TheEffingPhDHatersGonnaHate/PAM/defun_while.py in <module>()
51 t_x = tf.constant(np.random.randn(N,N),dtype = dType)
52 t_y = tf.constant(np.random.randn(N,N),dtype = dType)
---> 53 ys = _run(tf_test_log(t_x,t_y))
54
55
~/environments/tf_1_4_gpu/lib/python3.5/site-packages/tensorflow/python/framework/function.py in __call__(self, *args, **kwargs)
503
504 def __call__(self, *args, **kwargs):
--> 505 self.add_to_graph(ops.get_default_graph())
506 args = [ops.convert_to_tensor(_) for _ in args] + self._extra_inputs
507 ret, op = _call(self._signature, *args, **kwargs)
~/environments/tf_1_4_gpu/lib/python3.5/site-packages/tensorflow/python/framework/function.py in add_to_graph(self, g)
484 def add_to_graph(self, g):
485 """Adds this function into the graph g."""
--> 486 self._create_definition_if_needed()
487
488 # Adds this function into 'g'.
~/environments/tf_1_4_gpu/lib/python3.5/site-packages/tensorflow/python/framework/function.py in _create_definition_if_needed(self)
319 """Creates the function definition if it's not created yet."""
320 with context.graph_mode():
--> 321 self._create_definition_if_needed_impl()
322
323 def _create_definition_if_needed_impl(self):
~/environments/tf_1_4_gpu/lib/python3.5/site-packages/tensorflow/python/framework/function.py in _create_definition_if_needed_impl(self)
336 # Call func and gather the output tensors.
337 with vs.variable_scope("", custom_getter=temp_graph.getvar):
--> 338 outputs = self._func(*inputs)
339
340 # There is no way of distinguishing between a function not returning
~/Documents/TheEffingPhDHatersGonnaHate/PAM/defun_while.py in tf_test_log(t_x, t_y)
39 L,t_log_x = tf.while_loop(condition,body,ijk_0,
40 shape_invariants=[i0.get_shape(),
---> 41 tf.TensorShape([N,None])]
42 )
43 return t_log_x
~/environments/tf_1_4_gpu/lib/python3.5/site-packages/tensorflow/python/ops/control_flow_ops.py in while_loop(cond, body, loop_vars, shape_invariants, parallel_iterations, back_prop, swap_memory, name)
2814 loop_context = WhileContext(parallel_iterations, back_prop, swap_memory) # pylint: disable=redefined-outer-name
2815 ops.add_to_collection(ops.GraphKeys.WHILE_CONTEXT, loop_context)
-> 2816 result = loop_context.BuildLoop(cond, body, loop_vars, shape_invariants)
2817 return result
2818
~/environments/tf_1_4_gpu/lib/python3.5/site-packages/tensorflow/python/ops/control_flow_ops.py in BuildLoop(self, pred, body, loop_vars, shape_invariants)
2638 self.Enter()
2639 original_body_result, exit_vars = self._BuildLoop(
-> 2640 pred, body, original_loop_vars, loop_vars, shape_invariants)
2641 finally:
2642 self.Exit()
~/environments/tf_1_4_gpu/lib/python3.5/site-packages/tensorflow/python/ops/control_flow_ops.py in _BuildLoop(self, pred, body, original_loop_vars, loop_vars, shape_invariants)
2619 for m_var, n_var in zip(merge_vars, next_vars):
2620 if isinstance(m_var, ops.Tensor):
-> 2621 _EnforceShapeInvariant(m_var, n_var)
2622
2623 # Exit the loop.
~/environments/tf_1_4_gpu/lib/python3.5/site-packages/tensorflow/python/ops/control_flow_ops.py in _EnforceShapeInvariant(merge_var, next_var)
576 "Provide shape invariants using either the `shape_invariants` "
577 "argument of tf.while_loop or set_shape() on the loop variables."
--> 578 % (merge_var.name, m_shape, n_shape))
579 else:
580 if not isinstance(var, (ops.IndexedSlices, sparse_tensor.SparseTensor)):
ValueError: The shape for while/Merge_1:0 is not an invariant for the loop. It enters the loop with shape (100, ?), but has shape <unknown> after one iteration. Provide shape invariants using either the `shape_invariants` argument of tf.while_loop or set_shape() on the loop variables.
Thanks #Alexandre Passos for the suggestion in the comment above!
The following piece of code is a modification of the original with a set_shape function added inside the body.
import numpy as np
import tensorflow as tf
from tensorflow.python.framework import ops
from tensorflow.python.ops import array_ops
from tensorflow.python.ops import sparse_ops
from tensorflow.python.framework import function
def _run(tensor):
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
res = sess.run(tensor)
return res
#function.Defun(tf.float32,tf.float32,tf.float32,func_name ='tf_test_logGrad')
def tf_test_logGrad(t_x,t_y,grad):
return grad
#function.Defun(tf.float32,tf.float32,func_name ='tf_test_log')#,grad_func=tf_test_logGrad)
def tf_test_log(t_x,t_y):
#N = t_x.shape[0].value
condition = lambda i,m1: i<N
def body(index,x):
#return[(index+1),tf.concat([x, tf.expand_dims(tf.exp( tf.add( t_x[:,index],t_y[:,index]) ),1) ],1 ) ]
x = tf.add(x, tf.exp( tf.add( t_x[:,0],t_y[:,0]) ) )
x.set_shape([N])
return[(index+1), x]
i0 = tf.constant(0,dtype=tf.int32)
m0 = tf.zeros([N],dType)
ijk_0 = [i0,m0]
L,t_log_x = tf.while_loop(condition,body,ijk_0,
shape_invariants=[i0.get_shape(),
tf.TensorShape([N])]
)
return t_log_x
dType = tf.float32
N = np.int32(100)
t_N = tf.constant(N,dtype = tf.int32)
t_x = tf.constant(np.random.randn(N,N),dtype = dType)
t_y = tf.constant(np.random.randn(N,N),dtype = dType)
ys = _run(tf_test_log(t_x,t_y))
The Issue of global N still persists.
You still need to set the shape of the loop tensors as a global variable outside of the defun decorator. If you try to get it from the shape of the inputs of the defun decorator, you get:
TypeError Traceback (most recent call last)
~/environments/tf_1_4_gpu/lib/python3.5/site-packages/tensorflow/python/ops/array_ops.py in zeros(shape, dtype, name)
1438 shape = tensor_shape.as_shape(shape)
-> 1439 output = constant(zero, shape=shape, dtype=dtype, name=name)
1440 except (TypeError, ValueError):
~/environments/tf_1_4_gpu/lib/python3.5/site-packages/tensorflow/python/framework/constant_op.py in constant(value, dtype, shape, name, verify_shape)
207 tensor_util.make_tensor_proto(
--> 208 value, dtype=dtype, shape=shape, verify_shape=verify_shape))
209 dtype_value = attr_value_pb2.AttrValue(type=tensor_value.tensor.dtype)
~/environments/tf_1_4_gpu/lib/python3.5/site-packages/tensorflow/python/framework/tensor_util.py in make_tensor_proto(values, dtype, shape, verify_shape)
379 # exception when dtype is set to np.int64
--> 380 if shape is not None and np.prod(shape, dtype=np.int64) == 0:
381 nparray = np.empty(shape, dtype=np_dt)
~/environments/tf_1_4_gpu/lib/python3.5/site-packages/numpy/core/fromnumeric.py in prod(a, axis, dtype, out, keepdims)
2517 return _methods._prod(a, axis=axis, dtype=dtype,
-> 2518 out=out, **kwargs)
2519
~/environments/tf_1_4_gpu/lib/python3.5/site-packages/numpy/core/_methods.py in _prod(a, axis, dtype, out, keepdims)
34 def _prod(a, axis=None, dtype=None, out=None, keepdims=False):
---> 35 return umr_prod(a, axis, dtype, out, keepdims)
36
TypeError: __int__ returned non-int (type NoneType)
During handling of the above exception, another exception occurred:
ValueError Traceback (most recent call last)
~/Documents/TheEffingPhDHatersGonnaHate/PAM/defun_while.py in <module>()
52 t_x = tf.constant(np.random.randn(N,N),dtype = dType)
53 t_y = tf.constant(np.random.randn(N,N),dtype = dType)
---> 54 ys = _run(tf_test_log(t_x,t_y))
55
56
~/environments/tf_1_4_gpu/lib/python3.5/site-packages/tensorflow/python/framework/function.py in __call__(self, *args, **kwargs)
503
504 def __call__(self, *args, **kwargs):
--> 505 self.add_to_graph(ops.get_default_graph())
506 args = [ops.convert_to_tensor(_) for _ in args] + self._extra_inputs
507 ret, op = _call(self._signature, *args, **kwargs)
~/environments/tf_1_4_gpu/lib/python3.5/site-packages/tensorflow/python/framework/function.py in add_to_graph(self, g)
484 def add_to_graph(self, g):
485 """Adds this function into the graph g."""
--> 486 self._create_definition_if_needed()
487
488 # Adds this function into 'g'.
~/environments/tf_1_4_gpu/lib/python3.5/site-packages/tensorflow/python/framework/function.py in _create_definition_if_needed(self)
319 """Creates the function definition if it's not created yet."""
320 with context.graph_mode():
--> 321 self._create_definition_if_needed_impl()
322
323 def _create_definition_if_needed_impl(self):
~/environments/tf_1_4_gpu/lib/python3.5/site-packages/tensorflow/python/framework/function.py in _create_definition_if_needed_impl(self)
336 # Call func and gather the output tensors.
337 with vs.variable_scope("", custom_getter=temp_graph.getvar):
--> 338 outputs = self._func(*inputs)
339
340 # There is no way of distinguishing between a function not returning
~/Documents/TheEffingPhDHatersGonnaHate/PAM/defun_while.py in tf_test_log(t_x, t_y)
33
34 i0 = tf.constant(0,dtype=tf.int32)
---> 35 m0 = tf.zeros([N],dType)
36
37
~/environments/tf_1_4_gpu/lib/python3.5/site-packages/tensorflow/python/ops/array_ops.py in zeros(shape, dtype, name)
1439 output = constant(zero, shape=shape, dtype=dtype, name=name)
1440 except (TypeError, ValueError):
-> 1441 shape = ops.convert_to_tensor(shape, dtype=dtypes.int32, name="shape")
1442 output = fill(shape, constant(zero, dtype=dtype), name=name)
1443 assert output.dtype.base_dtype == dtype
~/environments/tf_1_4_gpu/lib/python3.5/site-packages/tensorflow/python/framework/ops.py in convert_to_tensor(value, dtype, name, preferred_dtype)
834 name=name,
835 preferred_dtype=preferred_dtype,
--> 836 as_ref=False)
837
838
~/environments/tf_1_4_gpu/lib/python3.5/site-packages/tensorflow/python/framework/ops.py in internal_convert_to_tensor(value, dtype, name, as_ref, preferred_dtype, ctx)
924
925 if ret is None:
--> 926 ret = conversion_func(value, dtype=dtype, name=name, as_ref=as_ref)
927
928 if ret is NotImplemented:
~/environments/tf_1_4_gpu/lib/python3.5/site-packages/tensorflow/python/framework/constant_op.py in _tensor_shape_tensor_conversion_function(s, dtype, name, as_ref)
248 if not s.is_fully_defined():
249 raise ValueError(
--> 250 "Cannot convert a partially known TensorShape to a Tensor: %s" % s)
251 s_list = s.as_list()
252 int64_value = 0
ValueError: Cannot convert a partially known TensorShape to a Tensor: (?,)
I'm trying to do a weighted Ridge Regression with sklearn. However, the code breaks when I call the fit method. The exception I get is :
Exception: Data must be 1-dimensional
But I'm sure (by checking through print-statements) that the data I'm passing has the right shapes.
print temp1.shape #(781, 21)
print temp2.shape #(781,)
print weights.shape #(781,)
result=RidgeCV(normalize=True).fit(temp1,temp2,sample_weight=weights)
What could be going wrong ??
Here's the whole output :
---------------------------------------------------------------------------
Exception Traceback (most recent call last)
<ipython-input-65-a5b1eba5d9cf> in <module>()
22
23
---> 24 result=RidgeCV(normalize=True).fit(temp2,temp1, sample_weight=weights)
25
26
/usr/local/lib/python2.7/dist-packages/sklearn/linear_model/ridge.pyc in fit(self, X, y, sample_weight)
868 gcv_mode=self.gcv_mode,
869 store_cv_values=self.store_cv_values)
--> 870 estimator.fit(X, y, sample_weight=sample_weight)
871 self.alpha_ = estimator.alpha_
872 if self.store_cv_values:
/usr/local/lib/python2.7/dist-packages/sklearn/linear_model/ridge.pyc in fit(self, X, y, sample_weight)
793 else alpha)
794 if error:
--> 795 out, c = _errors(weighted_alpha, y, v, Q, QT_y)
796 else:
797 out, c = _values(weighted_alpha, y, v, Q, QT_y)
/usr/local/lib/python2.7/dist-packages/sklearn/linear_model/ridge.pyc in _errors(self, alpha, y, v, Q, QT_y)
685 w = 1.0 / (v + alpha)
686 c = np.dot(Q, self._diag_dot(w, QT_y))
--> 687 G_diag = self._decomp_diag(w, Q)
688 # handle case where y is 2-d
689 if len(y.shape) != 1:
/usr/local/lib/python2.7/dist-packages/sklearn/linear_model/ridge.pyc in _decomp_diag(self, v_prime, Q)
672 def _decomp_diag(self, v_prime, Q):
673 # compute diagonal of the matrix: dot(Q, dot(diag(v_prime), Q^T))
--> 674 return (v_prime * Q ** 2).sum(axis=-1)
675
676 def _diag_dot(self, D, B):
/usr/local/lib/python2.7/dist-packages/pandas/core/ops.pyc in wrapper(left, right, name)
531 return left._constructor(wrap_results(na_op(lvalues, rvalues)),
532 index=left.index, name=left.name,
--> 533 dtype=dtype)
534 return wrapper
535
/usr/local/lib/python2.7/dist-packages/pandas/core/series.pyc in __init__(self, data, index, dtype, name, copy, fastpath)
209 else:
210 data = _sanitize_array(data, index, dtype, copy,
--> 211 raise_cast_failure=True)
212
213 data = SingleBlockManager(data, index, fastpath=True)
/usr/local/lib/python2.7/dist-packages/pandas/core/series.pyc in _sanitize_array(data, index, dtype, copy, raise_cast_failure)
2683 elif subarr.ndim > 1:
2684 if isinstance(data, np.ndarray):
-> 2685 raise Exception('Data must be 1-dimensional')
2686 else:
2687 subarr = _asarray_tuplesafe(data, dtype=dtype)
Exception: Data must be 1-dimensional
The error seems to be due to sample_weights being a Pandas series rather than a numpy array:
from sklearn.linear_model import RidgeCV
temp1 = pd.DataFrame(np.random.rand(781, 21))
temp2 = pd.Series(temp1.sum(1))
weights = pd.Series(1 + 0.1 * np.random.rand(781))
result = RidgeCV(normalize=True).fit(temp1, temp2,
sample_weight=weights)
# Exception: Data must be 1-dimensional
If you use a numpy array instead, the error goes away:
result = RidgeCV(normalize=True).fit(temp1, temp2,
sample_weight=weights.values)
This seems to be a bug; I've opened a scikit-learn issue to report this.