pip install pandas scikit-learn joblib numpy

import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestRegressor
from sklearn.metrics import r2_score, mean_squared_error
import joblib
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.preprocessing import LabelEncoder
import numpy as np# 1. 加载数据
data = pd.read_csv('amzondata.csv')print("数据预览：")
print(data.head())
print("\n数据集列名：")
print(data.columns.tolist())
print("\n数据类型：")
print(data.dtypes)# 2. 明确指定目标变量和主要特征
# 目标变量
TARGET_COL = 'final_price'
# 主要特征
MAIN_FEATURES = ['initial_price', 'rating']# 3. 数据清洗：只要求目标变量和主要数值特征非空
clean_cols = [TARGET_COL] + MAIN_FEATURESdata_clean = data.dropna(subset=clean_cols, how='any').reset_index(drop=True)
print(f"清洗后剩余样本数: {len(data_clean)}")
if len(data_clean) == 0:print("❌ 清洗后没有可用样本，请检查 amzondata.csv 的final_price、initial_price、rating三列是否有有效数据！")exit(1)X = data_clean.drop(TARGET_COL, axis=1)
y = data_clean[TARGET_COL]# 4. 特征处理
# 自动识别文本描述列（如包含desc、描述、description等）
desc_columns = [col for col in X.columns if any(k in col.lower() for k in ['description', 'desc', '描述'])]
other_object_columns = [col for col in X.select_dtypes(include=['object']).columns if col not in desc_columns]
numeric_columns = X.select_dtypes(include=['number']).columns.tolist()print(f"\n数值列: {numeric_columns}")
print(f"文本描述列: {desc_columns}")
print(f"其它分类型列: {other_object_columns}")# 处理文本描述列（TF-IDF）
tfidf_vectorizers = {}
desc_features = []
if desc_columns:for col in desc_columns:# 检查该列是否全为空或全是空白if X[col].dropna().astype(str).str.strip().replace('', pd.NA).dropna().empty:print(f"列 '{col}' 全为空或无有效文本，已跳过TF-IDF处理")continuetry:tfidf = TfidfVectorizer(max_features=50)desc_matrix = tfidf.fit_transform(X[col].astype(str)).toarray()desc_feature_names = [f"{col}_tfidf_{i}" for i in range(desc_matrix.shape[1])]desc_df = pd.DataFrame(desc_matrix, columns=desc_feature_names)desc_features.append(desc_df)tfidf_vectorizers[col] = tfidfexcept ValueError as e:print(f"列 '{col}' TF-IDF处理失败: {e}，已跳过")X = X.drop([col for col in desc_columns if col in X.columns], axis=1)if desc_features:desc_features = pd.concat(desc_features, axis=1)else:desc_features = pd.DataFrame()
else:desc_features = pd.DataFrame()# 处理其它object列（LabelEncoder）
encoders = {}
if other_object_columns:for col in other_object_columns:le = LabelEncoder()X[col] = le.fit_transform(X[col].astype(str))encoders[col] = le# 合并所有特征
data_features = [X.reset_index(drop=True)]
if not desc_features.empty:data_features.append(desc_features.reset_index(drop=True))
X_all = pd.concat(data_features, axis=1)# 检查并填充缺失值
if X_all.isnull().sum().sum() > 0:X_all = X_all.fillna(X_all.mean(numeric_only=True))print("已填充缺失值")# 检查样本数
if X_all.shape[0] == 0 or y.shape[0] == 0:print("❌ 预处理后没有可用样本，请检查数据源，确保final_price、initial_price、rating三列有有效数据。")exit(1)# 5. 划分训练集和测试集
X_train, X_test, y_train, y_test = train_test_split(X_all, y, test_size=0.2, random_state=42)# 6. 创建并训练回归模型
model = RandomForestRegressor(n_estimators=100, random_state=42)
model.fit(X_train, y_train)y_pred = model.predict(X_test)
r2 = r2_score(y_test, y_pred)
mse = mean_squared_error(y_test, y_pred)
print(f'模型R2分数: {r2:.4f}')
print(f'均方误差MSE: {mse:.2f}')results = pd.DataFrame({'实际值': y_test, '预测值': y_pred})
print("\n预测结果与实际结果对比：")
print(results.head())# 7. 保存模型、编码器、特征名、TF-IDF向量器
joblib.dump(model, 'trained_model.pkl')
joblib.dump(encoders, 'encoders.pkl')
joblib.dump(tfidf_vectorizers, 'tfidf_vectorizers.pkl')
feature_names = X_all.columns.tolist()
joblib.dump(feature_names, 'feature_names.pkl')
print(f"已保存模型、编码器、TF-IDF向量器和特征名称: {feature_names}")

python3 amzon.py

import pandas as pd# 1. 读取数据
data = pd.read_csv('amzondata.csv')# 明确指定主要列名
price_col = 'final_price'      # 价钱列
rating_col = 'rating'          # 评分列
desc_col = 'title'             # 商品名称/描述列# 检查列是否存在
for col in [price_col, rating_col, desc_col]:if col not in data.columns:raise ValueError(f'未找到列: {col}')# 去除主要列为空的行
data = data.dropna(subset=[price_col, rating_col, desc_col], how='any').reset_index(drop=True)# 2. 用户输入价钱
try:input_price = float(input(f"请输入你想要的相机价钱（如1000）：\n"))
except Exception:print("输入无效，请输入数字！")exit(1)# 3. 设定价钱区间（±10%）
price_range = 0.1
min_price = input_price * (1 - price_range)
max_price = input_price * (1 + price_range)# 4. 筛选
filtered = data[(data[price_col] >= min_price) & (data[price_col] <= max_price)]# 5. 输出结果
if filtered.empty:print("没有找到该价位附近的相机。")
else:print(f"在{min_price:.2f}~{max_price:.2f}元区间的相机：")# 截断title为前30字符display_df = filtered[[desc_col, price_col, rating_col]].copy()display_df[desc_col] = display_df[desc_col].astype(str).apply(lambda x: x[:30] + '...' if len(x) > 30 else x)print(display_df.sort_values(by=price_col).head(20).to_string(index=False))

python3 recommend_camera.py