Calculate MAE, MSE and R2 metrics for a DNNRegressor model - tensorflow

Question

I have a DNNRegressor model and I want to calculate some metrics to understand how well my model is predicting. How can I calculate the mean absolute error (MAE), mean squared error (MSE) and the R squared coefficient?
So far I only have the loss so can someone help me calculate MAE, MSE and R2?
# Imports
import itertools
import pandas as pd
import tensorflow as tf
import numpy as np
import sklearn
from sklearn import preprocessing
from sklearn.model_selection import train_test_split
from sklearn.metrics import r2_score
import seaborn as sns
from matplotlib import pyplot as plt
from sklearn.metrics import r2_score
import statsmodels.api as sm
COLUMNS = ['Prot', 'Gra', 'Cen', 'Sal', 'TVN', 'Velocidad_Prensa']
FEATURES = ['Prot', 'Gra', 'Cen', 'Sal', 'TVN']
LABEL = ['Velocidad_Prensa']
def get_input_fn(data_set, num_epochs=None, shuffle=True):
return tf.compat.v1.estimator.inputs.pandas_input_fn(
x=pd.DataFrame({k: data_set[k].values for k in FEATURES}),
y=pd.Series(data_set[LABEL].values),
num_epochs=num_epochs,
shuffle=shuffle)
training_set = pd.read_csv("prensa train.csv", skipinitialspace=True, skiprows=1, names=COLUMNS)
test_set = pd.read_csv("prensa eval.csv", skipinitialspace=True, skiprows=1, names=COLUMNS)
training_set.head()
# Model
feature_cols = [tf.feature_column.numeric_column(k) for k in FEATURES]
regressor = tf.estimator.DNNRegressor(feature_columns=feature_cols,
activation_fn = tf.nn.relu, hidden_units=[200, 100, 50, 25, 12])
# Reset the index of training
training_set.reset_index(drop = True, inplace =True)
def input_fn(data_set, pred = False):
if pred == False:
feature_cols = {k: tf.constant(data_set[k].values) for k in FEATURES}
labels = tf.constant(data_set[LABEL].values)
return feature_cols, labels
if pred == True:
feature_cols = {k: tf.constant(data_set[k].values) for k in FEATURES}
return feature_cols
# Deep Neural Network Regressor with the training set which contain the data split by train test split
regressor.train(input_fn=lambda: input_fn(training_set), steps=2000)
# Evaluation on the test set created by train_test_split
ev = regressor.evaluate(input_fn=lambda: input_fn(test_set), steps=1)
# Display the score on the testing set
loss_score1 = ev["loss"]
print("Final Loss on the testing set: {0:f}".format(loss_score1))
def input_fn(features, batch_size=256):
return tf.data.Dataset.from_tensor_slices(dict(features)).batch(batch_size)
features = ['Prot', 'Gra', 'Cen', 'Sal','TVN']
predict = {}
print("Ingresar características químicas de la materia prima")
for feature in features:
valid = True
while valid:
val = input(feature + ": ")
if not val.isdigit(): valid = False
predict[feature] = [float(val)]
predictions = regressor.predict(input_fn=lambda: input_fn(predict))
for pred_dict in predictions:
print(pred_dict)

Answer 1

sklearn.metrics has dedicated scoring methods for each of the metrics you are asking for.
Just to the following:
# Import metrics
from sklearn import metrics
# Make predictions
predictions = regressor.predict(input_fn=lambda: input_fn(predict))
# Calculate MAE, MSE, R2
print('MAE:', metrics.mean_absolute_error(y_true, predictions))
print('MSE:', metrics.mean_squared_error(y_true, predictions))
print('R2:', metrics.r2_score(y_true, predictions))