Chapter 7 – Feature Columns
Here’s my source code for the Feature Columns example in Chapter 7 of O’Reilly’s book “Learning Tensorflow”.
|
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 |
import tensorflow as tf import numpy as np import pandas as pd from tensorflow.contrib import learn N = 10000 weight = np.random.randn(N)*5+70 spec_id = np.random.randint(0,3,N) bias = [0.9,1,1.1] height = np.array([weight[i]/100 + bias[b] for i,b in enumerate(spec_id)]) spec_name = ['Goblin', 'Human', 'Manbears'] spec = [spec_name[s] for s in spec_id] df = pd.DataFrame({'Species':spec, 'Weight':weight, 'Height':height}) from tensorflow.contrib import layers Weight = layers.real_valued_column("Weight") Species = layers.sparse_column_with_keys(column_name="Species", keys=['Goblin', 'Human', 'Manbears']) # reg = learn.LinearRegressor(feature_columns=[Weight, Species]) reg = tf.estimator.LinearRegressor(feature_columns=[Weight, Species]) def input_fn(df): feature_cols = {} feature_cols['Weight'] = tf.constant(df['Weight'].values) feature_cols['Species'] = tf.SparseTensor(indices=[[i, 0] for i in range(df['Species'].size)], values=df['Species'].values, dense_shape=[df['Species'].size, 1]) labels = tf.constant(df['Height'].values) return feature_cols, labels # SparseTensor(indices=[[0, 0], [2, 1], [2, 2]], values=[2, 5, 7], dense_shape=[3, 3]) # [0, 0] -> in the first row, put the first element # [2, 1] -> in the third row, put the second element # [2, 2] -> in the third row, put the third element # reg.fit(input_fn=lambda:input_fn(df), steps=500) reg.train(input_fn=lambda:input_fn(df), steps=500) # print(reg.get_variable_names()) # w_w = reg.get_variable_value('linear/Weight/weight') w_w = reg.get_variable_value('linear/linear_model/Weight/weights') print('Estimation for Weight: {}'.format(w_w)) # s_w = reg.get_variable_value('linear/Species/weights') s_w = reg.get_variable_value('linear/linear_model/Species/weights') # b = reg.get_variable_value('linear/bias_weight') b = reg.get_variable_value('linear/linear_model/bias_weights') print('Estimation for Species: {}'.format(s_w + b)) |
As of TF 1.3 the code they provided was deprecated (not their fault, things change). They leave out some of the imports (namely Tensorflow, numpy, and learn from tensorflow.contrib, although the latter is negated when we correct […]
