Tensorflow : Creating a loop for feature variables

Question

I have multiple feature columns in TensorFlow and I am trying to create the loop to avoid manual typing to create feature column but its not working. Below is the list of all the columns which I need to create feature columns(this is just a dummy data, but what if we have hundreds of columns).

num_preg = tf.feature_column.numeric_column('Number_pregnant')
glucose_conc = tf.feature_column.numeric_column('glucose_concentration')
blood_prs = tf.feature_column.numeric_column('blood_pressure')
Tricep = tf.feature_column.numeric_column('Triceps')
insulin = tf.feature_column.numeric_column('Insulin')

I created a loop for this as mentioned below, df_col_num is a list containing the column names for which I need to create feature column.

for col in df_col_num:
    col= tf.feature_column.categorical_column_with_hash_bucket(col,hash_bucke[enter link description here][1]t_size=50)

Any help or suggestions would be appreciated.

Thanks !

see stackoverflow.com/questions/46834680/…. Basically if you can load your data through pandas (also you can apply filters too), you iterate through all columns, adding each to the list of features, without the need to declare each of them explicitly. — greeness
– greeness, Commented Nov 21, 2017 at 1:38

milo526 · Accepted Answer · 2018-01-24 10:16:12Z

3

your variables are numeric, but you are using tf.feature_column.categorical_column_with_hash_bucket, which is for categorical variables.

I'd use:

variable_names = df.columns

tf_variables =[]

for index in variable_names:
    tf_variables.append(tf.feature_column.numeric_column(index))

This worked for me.

edited Jan 24, 2018 at 10:16

milo526

5,1035 gold badges47 silver badges63 bronze badges

answered Jan 24, 2018 at 10:11

koe82

312 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

Tensorflow : Creating a loop for feature variables

1 Answer 1

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related