ValueError: Features are incompatible with given information - tensorflow

Question

While trying to follow the sample code for TensorFlow Linear Model Tutorial
No change to the structure of the code, only the columns are changed.
while running, i encounter the error"ValueError: Features are incompatible with given information."
Given Features:
'COLUMN_1': <tensorflow.python.framework.sparse_tensor.SparseTensor object at 0x00000000094C2D30>
'COLUMN_2': <tensorflow.python.framework.sparse_tensor.SparseTensor object at 0x000000000C4D4B38>
'COLUMN_3': <tf.Tensor 'Const_1:0' shape=(3,) dtype=float64>
required signatures::
'COLUMN_1': TensorSignature(dtype=tf.string, shape=None, is_sparse=True)
'COLUMN_2': TensorSignature(dtype=tf.string, shape=None, is_sparse=True)
'COLUMN_3': TensorSignature(dtype=tf.float64, shape=TensorShape([Dimension(3)]), is_sparse=False)
issue is that the test data and the training data are processed by the same function, the output structure should be the same.
i did testing to use training data as test data, it works,no error.
so how does the data in test set affects the feature signature?
same error was asked here
Full error message
traceback (most recent call last):
File "C:\Users\USERA\Desktop\USERA\New Projects 2017\Machine Learning\Renewal\TensorFlowLinearModelRenewal.py", line 308, in <module>
tf.app.run(main=main, argv=[sys.argv[0]] + unparsed)
File "C:\Users\USERA\AppData\Local\Programs\Python\Python35\lib\site-packages\tensorflow\python\platform\app.py", line 48, in run
_sys.exit(main(_sys.argv[:1] + flags_passthrough))
File "C:\Users\USERA\Desktop\USERA\New Projects 2017\Machine Learning\Renewal\TensorFlowLinearModelRenewal.py", line 270, in main
FLAGS.train_data, FLAGS.test_data)
File "C:\Users\USERA\Desktop\USERA\New Projects 2017\Machine Learning\Renewal\TensorFlowLinearModelRenewal.py", line 255, in train_and_eval
results = m.evaluate(input_fn=lambda: input_fn(df_test), steps=1)
File "C:\Users\USERA\AppData\Local\Programs\Python\Python35\lib\site-packages\tensorflow\python\util\deprecation.py", line 289, in new_func
return func(*args, **kwargs)
File "C:\Users\USERA\AppData\Local\Programs\Python\Python35\lib\site-packages\tensorflow\contrib\learn\python\learn\estimators\estimator.py", line 543, in evaluate
log_progress=log_progress)
File "C:\Users\USERA\AppData\Local\Programs\Python\Python35\lib\site-packages\tensorflow\contrib\learn\python\learn\estimators\estimator.py", line 827, in _evaluate_model
self._check_inputs(features, labels)
File "C:\Users\USERA\AppData\Local\Programs\Python\Python35\lib\site-packages\tensorflow\contrib\learn\python\learn\estimators\estimator.py", line 757, in _check_inputs
(str(features), str(self._features_info)))
ValueError: Features are incompatible with given information. Given features: {'POL_SRC_BUS_CODE': <tensorflow.python.framework.sparse_tensor.SparseTensor object at 0x00000000152A2320>, 'POL_SUB_PRD_CODE': <tensorflow.python.framework.sparse_tensor.SparseTensor object at 0x00000000120DE080>, 'ASSR_TYPE': <tensorflow.python.framework.sparse_tensor.SparseTensor object at 0x00000000152A2518>, 'POL_PYMT_MODE': <tensorflow.python.framework.sparse_tensor.SparseTensor object at 0x000000000E523FD0>, 'POL_PREM_RENEWAL': <tf.Tensor 'Const_2:0' shape=(12,) dtype=float64>, 'POL_JACKET_CODE': <tensorflow.python.framework.sparse_tensor.SparseTensor object at 0x0000000015D4C7F0>, 'POL_AGE': <tf.Tensor 'Const:0' shape=(12,) dtype=int64>, 'ASSR_GENDER': <tensorflow.python.framework.sparse_tensor.SparseTensor object at 0x000000000D92C2B0>, 'ACCOUNT_CLASS': <tensorflow.python.framework.sparse_tensor.SparseTensor object at 0x000000000B40FE80>, 'POL_PREM_ORIGINAL': <tf.Tensor 'Const_1:0' shape=(12,) dtype=float64>, 'POL_SCHEME_PLAN': <tensorflow.python.framework.sparse_tensor.SparseTensor object at 0x00000000152A2908>, 'POL_CUST_CODE': <tensorflow.python.framework.sparse_tensor.SparseTensor object at 0x000000000BCF27F0>, 'CLAIM_INCURRED': <tf.Tensor 'Const_4:0' shape=(12,) dtype=float64>, 'POL_END_NO_IDX': <tf.Tensor 'Const_3:0' shape=(12,) dtype=int64>, 'ASSR_MAR_STATUS': <tensorflow.python.framework.sparse_tensor.SparseTensor object at 0x00000000152A2208>, 'POL_OCC_DESC': <tensorflow.python.framework.sparse_tensor.SparseTensor object at 0x00000000152A27F0>, 'ASSR_NATIONALITY': <tensorflow.python.framework.sparse_tensor.SparseTensor object at 0x000000000DB4F588>}, required signatures: {'ASSR_GENDER': TensorSignature(dtype=tf.string, shape=None, is_sparse=True), 'POL_PREM_RENEWAL': TensorSignature(dtype=tf.float64, shape=TensorShape([Dimension(21)]), is_sparse=False), 'POL_PREM_ORIGINAL': TensorSignature(dtype=tf.float64, shape=TensorShape([Dimension(21)]), is_sparse=False), 'POL_SUB_PRD_CODE': TensorSignature(dtype=tf.string, shape=None, is_sparse=True), 'ASSR_NATIONALITY': TensorSignature(dtype=tf.string, shape=None, is_sparse=True), 'POL_CUST_CODE': TensorSignature(dtype=tf.string, shape=None, is_sparse=True), 'POL_SRC_BUS_CODE': TensorSignature(dtype=tf.string, shape=None, is_sparse=True), 'ASSR_MAR_STATUS': TensorSignature(dtype=tf.string, shape=None, is_sparse=True), 'CLAIM_INCURRED': TensorSignature(dtype=tf.int64, shape=TensorShape([Dimension(21)]), is_sparse=False), 'POL_END_NO_IDX': TensorSignature(dtype=tf.int64, shape=TensorShape([Dimension(21)]), is_sparse=False), 'POL_SCHEME_PLAN': TensorSignature(dtype=tf.string, shape=None, is_sparse=True), 'ASSR_TYPE': TensorSignature(dtype=tf.string, shape=None, is_sparse=True), 'POL_PYMT_MODE': TensorSignature(dtype=tf.string, shape=None, is_sparse=True), 'ACCOUNT_CLASS': TensorSignature(dtype=tf.string, shape=None, is_sparse=True), 'POL_JACKET_CODE': TensorSignature(dtype=tf.string, shape=None, is_sparse=True), 'POL_OCC_DESC': TensorSignature(dtype=tf.string, shape=None, is_sparse=True), 'POL_AGE': TensorSignature(dtype=tf.int64, shape=TensorShape([Dimension(21)]), is_sparse=False)}.
my code
# Copyright 2016 The TensorFlow Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ==============================================================================
"""Example code for TensorFlow Wide & Deep Tutorial using TF.Learn API."""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import argparse
import sys
import tempfile
from six.moves import urllib
import pandas as pd
import tensorflow as tf
COLUMNS = ["EXPIRY_MONTH", "POL_SYS_ID","POL_DEPT_CODE","ACCOUNT_CLASS","POL_CUST_CODE","POL_SUB_PRD_CODE",
"POL_SRC_BUS_CODE","POL_OCC_DESC","POL_AGE","POL_INSURED_ID","ASSR_TYPE","ASSR_GENDER","ASSR_MAR_STATUS","ASSR_NATIONALITY",
"POL_SCHEME_PLAN","POL_PYMT_MODE","POL_POSTAL_CODE","POL_JACKET_CODE","POL_PREM_ORIGINAL","POL_PREM_RENEWAL","POL_END_NO_IDX",
"CLAIM_INCURRED","REVIEW_STATUS",]
LABEL_COLUMN = "label"
CATEGORICAL_COLUMNS = [ "ACCOUNT_CLASS", "POL_CUST_CODE", "POL_SUB_PRD_CODE",
"POL_SRC_BUS_CODE", "POL_OCC_DESC", "ASSR_TYPE", "ASSR_GENDER",
"ASSR_MAR_STATUS", "ASSR_NATIONALITY", "POL_SCHEME_PLAN", "POL_PYMT_MODE",
"POL_JACKET_CODE"
]
CONTINUOUS_COLUMNS = ["POL_AGE", "POL_PREM_ORIGINAL", "POL_PREM_RENEWAL","POL_END_NO_IDX","CLAIM_INCURRED"]
def maybe_download(train_data, test_data):
"""Maybe downloads training data and returns train and test file names."""
print('-----start of maybe_download')
if train_data:
train_file_name = train_data
else:
train_file = open('Renewal Listing 2015 export2.csv')
train_file_name = train_file.name
train_file.close()
print("------maybe_download()-----Training data is downloaded to %s" % train_file_name)
if test_data:
test_file_name = test_data
else:
test_file = open('Renewal Listing 2015 export2 Test.csv')
test_file_name = test_file.name
test_file.close()
print("------maybe_download()-----Test data is downloaded to %s" % test_file_name)
print('-----end of maybe_download')
return train_file_name, test_file_name
def build_estimator(model_dir, model_type):
"""Build an estimator."""
print('-----start of build_estimator')
# Sparse base columns.
ACCOUNT_CLASS = tf.contrib.layers.sparse_column_with_hash_bucket(
"ACCOUNT_CLASS", hash_bucket_size=1000)
POL_CUST_CODE = tf.contrib.layers.sparse_column_with_hash_bucket(
"POL_CUST_CODE", hash_bucket_size=100)
POL_SUB_PRD_CODE = tf.contrib.layers.sparse_column_with_hash_bucket(
"POL_SUB_PRD_CODE", hash_bucket_size=100)
POL_SRC_BUS_CODE = tf.contrib.layers.sparse_column_with_hash_bucket(
"POL_SRC_BUS_CODE", hash_bucket_size=1000)
POL_OCC_DESC = tf.contrib.layers.sparse_column_with_hash_bucket(
"POL_OCC_DESC", hash_bucket_size=1000)
ASSR_NATIONALITY = tf.contrib.layers.sparse_column_with_hash_bucket(
"ASSR_NATIONALITY", hash_bucket_size=1000)
POL_JACKET_CODE = tf.contrib.layers.sparse_column_with_hash_bucket(
"POL_JACKET_CODE", hash_bucket_size=1000)
print("----build_estimator()----hash_bucket_size: columns are processed")
ASSR_GENDER = tf.contrib.layers.sparse_column_with_keys(column_name="ASSR_GENDER",
keys=["M", "F"])
ASSR_TYPE = tf.contrib.layers.sparse_column_with_keys(column_name="ASSR_TYPE",
keys=["I", "C","M"])
ASSR_MAR_STATUS = tf.contrib.layers.sparse_column_with_keys(column_name="ASSR_MAR_STATUS",
keys=["D", "M","S","W"])
POL_SCHEME_PLAN = tf.contrib.layers.sparse_column_with_keys(column_name="POL_SCHEME_PLAN",
keys=["Y", "N"])
POL_PYMT_MODE = tf.contrib.layers.sparse_column_with_keys(column_name="POL_PYMT_MODE",
keys=["C", "CH","CR"])
print("----build_estimator()----sparse_column_with_keys: columns are processed")
# Continuous base columns.
POL_AGE = tf.contrib.layers.real_valued_column("POL_AGE")
POL_PREM_ORIGINAL = tf.contrib.layers.real_valued_column("POL_PREM_ORIGINAL")
POL_PREM_RENEWAL = tf.contrib.layers.real_valued_column("POL_PREM_RENEWAL")
POL_END_NO_IDX = tf.contrib.layers.real_valued_column("POL_END_NO_IDX")
CLAIM_INCURRED = tf.contrib.layers.real_valued_column("CLAIM_INCURRED")
print("----build_estimator()----Continuous base columns; are processed")
# Transformations.
age_buckets = tf.contrib.layers.bucketized_column(POL_AGE,
boundaries=[
18, 25, 30, 35, 40, 45,
50, 55, 60, 65
])
print("----build_estimator()----Transformations for age to boundaries is processed")
# Wide columns and deep columns.
wide_columns = [ACCOUNT_CLASS, POL_CUST_CODE, POL_SUB_PRD_CODE, POL_SRC_BUS_CODE, POL_OCC_DESC,
ASSR_NATIONALITY, POL_JACKET_CODE,ASSR_GENDER,ASSR_TYPE,ASSR_MAR_STATUS,POL_SCHEME_PLAN,POL_PYMT_MODE,
POL_PREM_ORIGINAL,POL_PREM_RENEWAL,POL_END_NO_IDX,CLAIM_INCURRED,
age_buckets,
tf.contrib.layers.crossed_column([POL_CUST_CODE, POL_SUB_PRD_CODE],
hash_bucket_size=int(1e4)),
tf.contrib.layers.crossed_column(
[age_buckets, POL_OCC_DESC],
hash_bucket_size=int(1e4)),
tf.contrib.layers.crossed_column([ACCOUNT_CLASS, ASSR_TYPE],
hash_bucket_size=int(1e4))]
deep_columns = [
tf.contrib.layers.embedding_column(POL_OCC_DESC, dimension=8),
tf.contrib.layers.embedding_column(POL_SRC_BUS_CODE, dimension=8),
tf.contrib.layers.embedding_column(ASSR_GENDER, dimension=8),
tf.contrib.layers.embedding_column(POL_JACKET_CODE, dimension=8),
tf.contrib.layers.embedding_column(ASSR_NATIONALITY,
dimension=8),
tf.contrib.layers.embedding_column(POL_PYMT_MODE, dimension=8),
POL_AGE,
POL_PREM_ORIGINAL,
POL_PREM_RENEWAL,
POL_END_NO_IDX,
CLAIM_INCURRED,
]
if model_type == "wide":
m = tf.contrib.learn.LinearClassifier(model_dir=model_dir,
feature_columns=wide_columns)
elif model_type == "deep":
m = tf.contrib.learn.DNNClassifier(model_dir=model_dir,
feature_columns=deep_columns,
hidden_units=[100, 50])
else:
m = tf.contrib.learn.DNNLinearCombinedClassifier(
model_dir=model_dir,
linear_feature_columns=wide_columns,
dnn_feature_columns=deep_columns,
dnn_hidden_units=[100, 50],
fix_global_step_increment_bug=True)
print("-----end of build_estimator")
return m
def input_fn(df):
"""Input builder function."""
# Creates a dictionary mapping from each continuous feature column name (k) to
# the values of that column stored in a constant Tensor.
print('---------------------------------------------')
print('-----start of input_fn')
print('-----input_fn()----print df')
#print(df)
print('-----input_fn()----print df end')
continuous_cols = {k: tf.constant(df[k].values) for k in CONTINUOUS_COLUMNS}
print('-----input_fn()----print continuous_cols')
#print(continuous_cols)
print('-----input_fn()----print continuous_cols end')
# Creates a dictionary mapping from each categorical feature column name (k)
# to the values of that column stored in a tf.SparseTensor.
categorical_cols = {
k: tf.SparseTensor(
indices=[[i, 0] for i in range(df[k].size)],
values=df[k].values,
dense_shape=[df[k].size, 1])
for k in CATEGORICAL_COLUMNS}
print('-----input_fn()----print categorical_cols')
#print(categorical_cols)
print('-----input_fn()----print categorical_cols end')
# Merges the two dictionaries into one.
feature_cols = dict(continuous_cols)
feature_cols.update(categorical_cols)
print('-----input_fn()----print feature_cols')
#print(feature_cols)
print('-----input_fn()----end of print feature_cols')
# Converts the label column into a constant Tensor.
label = tf.constant(df[LABEL_COLUMN].values)
# Returns the feature columns and the label.
print('-----input_fn()----print label')
#print(label)
print('-----input_fn()----end of print label')
print('---------------------------------------------')
print('-----end of input_fn')
return feature_cols, label
def train_and_eval(model_dir, model_type, train_steps, train_data, test_data):
"""Train and evaluate the model."""
print('---------------------------------------------')
print('-----start of train_and_eval')
print("-----train_and_eval()-----start to get files")
train_file_name, test_file_name = maybe_download(train_data, test_data)
print("-----train_and_eval()-----got train and test data")
df_train = pd.read_csv(
tf.gfile.Open(train_file_name),
names=COLUMNS,
skipinitialspace=True,
engine="python")
df_test = pd.read_csv(
tf.gfile.Open(test_file_name),
names=COLUMNS,
skipinitialspace=True,
skiprows=1,
engine="python")
#print("----print df_train file")
#for row in df_train:
# print(row)
# remove NaN elements
df_train = df_train.dropna(how='any', axis=0)
df_test = df_test.dropna(how='any', axis=0)
df_train[LABEL_COLUMN] = (
df_train["REVIEW_STATUS"].apply(lambda x: "RENEWED" in x)).astype(int)
df_test[LABEL_COLUMN] = (
df_test["REVIEW_STATUS"].apply(lambda x: "RENEWED" in x)).astype(int)
model_dir = tempfile.mkdtemp() if not model_dir else model_dir
print("-----train_and_eval()-----model directory = %s" % model_dir)
m = build_estimator(model_dir, model_type)
print(m)
print('-----train_and_eval()-----finished build_estimator')
print('----------------------------------')
print('----------------------------------')
print('-----start of input_fn(df_train)-------------')
print('========== compare train and test')
print('train')
print(df_train.shape)
print(df_train['POL_JACKET_CODE'].shape)
print('test')
print(df_test.shape)
print(df_test)
print('=====================================')
m.fit(input_fn=lambda: input_fn(df_train), steps=train_steps)
print('-----end of input_fn(df_train)-------------')
print('-----beginning of evaluate input_fn(df_test)')
results = m.evaluate(input_fn=lambda: input_fn(df_test), steps=1)
print('-----train_and_eval()-----end of input_fn(df_test)')
print(' start to print -----results--------')
print(results)
for key in sorted(results):
print("%s: %s" % (key, results[key]))
print('---------------------------------------------')
print('-----end of train_and_eval')
FLAGS = None
def main(_):
print("-------main()----- program start")
train_and_eval(FLAGS.model_dir, FLAGS.model_type, FLAGS.train_steps,
FLAGS.train_data, FLAGS.test_data)
print("------main()----- program ended: ")
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.register("type", "bool", lambda v: v.lower() == "true")
parser.add_argument(
"--model_dir",
type=str,
default="",
help="Base directory for output models."
)
parser.add_argument(
"--model_type",
type=str,
default="wide_n_deep",
help="Valid model types: {'wide', 'deep', 'wide_n_deep'}."
)
parser.add_argument(
"--train_steps",
type=int,
default=200,
help="Number of training steps."
)
parser.add_argument(
"--train_data",
type=str,
default="",
help="Path to the training data."
)
parser.add_argument(
"--test_data",
type=str,
default="",
help="Path to the test data."
)
FLAGS, unparsed = parser.parse_known_args()
tf.app.run(main=main, argv=[sys.argv[0]] + unparsed)

Answer 1

i have found out the issue, when python read CSV, the data type depends on the data itself.
for me, there is a column with all data =0 in the test set, but its a float number column. when the program read test set csv, the column is considered as int. so it not matching with float number type in training data, thus the error. change 0 to 0.00 or use actual data where both 0 and floating number are in the column solved the issue.