计算 DNNRegressor 模型的 MAE、MSE 和 R2 指标

Question

我有一个DNNRegressor 模型，我想计算一些指标来了解我的模型的预测效果。如何计算平均绝对误差 (MAE)、均方误差 (MSE) 和 R 平方系数？

到目前为止，我只有损失，所以有人可以帮我计算 MAE、MSE 和 R2 吗？

# 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 对您要求的每个指标都有专门的评分方法。

就以下几点：

# 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))