使用tensorflow的线性回归的例子(九)

from future import absolute_import, division, print_function, unicode_literals

import numpy as np

import pandas as pd

import seaborn as sb

import tensorflow as tf

from tensorflow import keras as ks

from tensorflow.estimator import LinearRegressor

from sklearn import datasets

from sklearn.model_selection import train_test_split

from sklearn.metrics import mean_squared_error, r2_score

print(tf.version)

"""## Load and configure the Boston Housing Dataset"""

boston_load = datasets.load_boston()

feature_columns = boston_load.feature_names

target_column = boston_load.target

boston_data = pd.DataFrame(boston_load.data, columns=feature_columns).astype(np.float32)

boston_data['MEDV'] = target_column.astype(np.float32)

boston_data.head()

"""## Checking the relation between the variables using Pairplot and Correlation Graph"""

sb.pairplot(boston_data, diag_kind="kde", height=3, aspect=0.6)

correlation_data = boston_data.corr()

correlation_data.style.background_gradient(cmap='coolwarm', axis=None)

"""## Descriptive Statistics - Central Tendency and Dispersion"""

stats = boston_data.describe()

boston_stats = stats.transpose()

boston_stats

"""## Select the required columns"""

X_data = boston_data[[i for i in boston_data.columns if i not in ['MEDV']]]

Y_data = boston_data[['MEDV']]

"""## Train Test Split"""

training_features , test_features ,training_labels, test_labels = train_test_split(X_data , Y_data , test_size=0.2)

print('No. of rows in Training Features: ', training_features.shape[0])

print('No. of rows in Test Features: ', test_features.shape[0])

print('No. of columns in Training Features: ', training_features.shape[1])

print('No. of columns in Test Features: ', test_features.shape[1])

print('No. of rows in Training Label: ', training_labels.shape[0])

print('No. of rows in Test Label: ', test_labels.shape[0])

print('No. of columns in Training Label: ', training_labels.shape[1])

print('No. of columns in Test Label: ', test_labels.shape[1])

stats = training_features.describe()

stats = stats.transpose()

stats

stats = test_features.describe()

stats = stats.transpose()

stats

"""## Normalize Data"""

def norm(x):

stats = x.describe()

stats = stats.transpose()

return (x - stats['mean']) / stats['std']

normed_train_features = norm(training_features)

normed_test_features = norm(test_features)

"""## Build the Input Pipeline for TensorFlow model"""

def feed_input(features_dataframe, target_dataframe, num_of_epochs=10, shuffle=True, batch_size=32):

def input_feed_function():

dataset = tf.data.Dataset.from_tensor_slices((dict(features_dataframe), target_dataframe))

if shuffle:

dataset = dataset.shuffle(2000)

dataset = dataset.batch(batch_size).repeat(num_of_epochs)

return dataset

return input_feed_function

train_feed_input = feed_input(normed_train_features, training_labels)

train_feed_input_testing = feed_input(normed_train_features, training_labels, num_of_epochs=1, shuffle=False)

test_feed_input = feed_input(normed_test_features, test_labels, num_of_epochs=1, shuffle=False)

"""## Model Training"""

feature_columns_numeric = [tf.feature_column.numeric_column(m) for m in training_features.columns]

linear_model = LinearRegressor(feature_columns=feature_columns_numeric, optimizer='RMSProp')

linear_model.train(train_feed_input)

"""## Predictions"""

train_predictions = linear_model.predict(train_feed_input_testing)

test_predictions = linear_model.predict(test_feed_input)

train_predictions_series = pd.Series([p['predictions'][0] for p in train_predictions])

test_predictions_series = pd.Series([p['predictions'][0] for p in test_predictions])

train_predictions_df = pd.DataFrame(train_predictions_series, columns=['predictions'])

test_predictions_df = pd.DataFrame(test_predictions_series, columns=['predictions'])

training_labels.reset_index(drop=True, inplace=True)

train_predictions_df.reset_index(drop=True, inplace=True)

test_labels.reset_index(drop=True, inplace=True)

test_predictions_df.reset_index(drop=True, inplace=True)

train_labels_with_predictions_df = pd.concat([training_labels, train_predictions_df], axis=1)

test_labels_with_predictions_df = pd.concat([test_labels, test_predictions_df], axis=1)

"""## Validation"""

def calculate_errors_and_r2(y_true, y_pred):

mean_squared_err = (mean_squared_error(y_true, y_pred))

root_mean_squared_err = np.sqrt(mean_squared_err)

r2 = round(r2_score(y_true, y_pred)*100,0)

return mean_squared_err, root_mean_squared_err, r2

train_mean_squared_error, train_root_mean_squared_error, train_r2_score_percentage = calculate_errors_and_r2(training_labels, train_predictions_series)

test_mean_squared_error, test_root_mean_squared_error, test_r2_score_percentage = calculate_errors_and_r2(test_labels, test_predictions_series)

print('Training Data Mean Squared Error = ', train_mean_squared_error)

print('Training Data Root Mean Squared Error = ', train_root_mean_squared_error)

print('Training Data R2 = ', train_r2_score_percentage)

print('Test Data Mean Squared Error = ', test_mean_squared_error)

print('Test Data Root Mean Squared Error = ', test_root_mean_squared_error)

print('Test Data R2 = ', test_r2_score_percentage)

相关推荐
空中湖2 天前
PyTorch武侠演义 第一卷:初入江湖 第5章:玉如意的秘密
人工智能·pytorch·neo4j
cwn_2 天前
tensorflow搭建神经网络
人工智能·rnn·深度学习·神经网络·机器学习·cnn·tensorflow
文浩(楠搏万)6 天前
TensorFlow+CNN垃圾分类深度学习全流程实战教程
深度学习·ai·cnn·tensorflow·卷积神经网络·模型训练·全流程指导
少年攻城狮6 天前
Neo4j系列---【Linux离线安装neo4j】
linux·运维·服务器·neo4j
verify.Mar6 天前
docker && Neo4j
docker·容器·neo4j
Kakaxiii7 天前
【2025最新】使用neo4j实现GraphRAG所需的向量检索
自然语言处理·知识图谱·neo4j·向量数据库
BTU_YC8 天前
Neo4j Python 驱动库完整教程(带输入输出示例)
开发语言·python·neo4j
AI妈妈手把手8 天前
【深度学习框架终极PK】TensorFlow/PyTorch/MindSpore深度解析!选对框架效率翻倍
人工智能·pytorch·python·深度学习·tensorflow·mindspore·ai选型指南
曾几何时`9 天前
基于python和neo4j构建知识图谱医药问答系统
python·知识图谱·neo4j
Wendy14419 天前
【线性回归(最小二乘法MSE)】——机器学习
算法·机器学习·线性回归