📘 牛油果价格和销量数据可视化分析与预测/牛油果.ipynb

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
import plotly.express as px
import warnings
warnings.filterwarnings('ignore')
sns.set_style('darkgrid')
plt.style.use('ggplot')
avocado_data = pd.read_csv('Avocado_HassAvocadoBoard_20152023v1.0.1.csv')
avocado_data.head()

	Date	AveragePrice	TotalVolume	plu4046	plu4225	plu4770	TotalBags	SmallBags	LargeBags	XLargeBags	type	region
0	2015-01-04	1.22	40873.28	2819.50	28287.42	49.90	9716.46	9186.93	529.53	0.0	conventional	Albany
1	2015-01-04	1.79	1373.95	57.42	153.88	0.00	1162.65	1162.65	0.00	0.0	organic	Albany
2	2015-01-04	1.00	435021.49	364302.39	23821.16	82.15	46815.79	16707.15	30108.64	0.0	conventional	Atlanta
3	2015-01-04	1.76	3846.69	1500.15	938.35	0.00	1408.19	1071.35	336.84	0.0	organic	Atlanta
4	2015-01-04	1.08	788025.06	53987.31	552906.04	39995.03	141136.68	137146.07	3990.61	0.0	conventional	BaltimoreWashington

# 检查并计数重复的行
duplicate_count = avocado_data.duplicated().sum()
print(f"Number of duplicated rows: {duplicate_count}")

Number of duplicated rows: 0

# 根据“平均价格”列中的IQR识别异常值
Q1 = avocado_data['AveragePrice'].quantile(0.25)
Q3 = avocado_data['AveragePrice'].quantile(0.75)
IQR = Q3 - Q1
lower_bound = Q1 - 1.5 * IQR
upper_bound = Q3 + 1.5 * IQR
outliers_price = (avocado_data['AveragePrice'] < lower_bound) | (avocado_data['AveragePrice'] > upper_bound)
# 打印异常值的数量
print(f"Number of outliers in AveragePrice: {outliers_price.sum()}")

Number of outliers in AveragePrice: 358

avocado_data=avocado_data[~outliers_price]

# 识别有缺失值的行
missing_rows = avocado_data[avocado_data[['SmallBags', 'LargeBags', 'XLargeBags']].isnull().any(axis=1)]
# 遍历缺失的行
for index, row in missing_rows.iterrows():
    # 生成总和为1的随机百分比
    random_percentages = np.random.dirichlet(np.ones(3), size=1)[0]
    # 计算使总和等于TotalBags所需的剩余值
    remaining_value = row['TotalBags'] - row[['SmallBags', 'LargeBags', 'XLargeBags']].sum()
    # 用随机百分比填充缺失值
    avocado_data.at[index, 'SmallBags'] = remaining_value * random_percentages[0]
    avocado_data.at[index, 'LargeBags'] = remaining_value * random_percentages[1]
    avocado_data.at[index, 'XLargeBags'] = remaining_value * random_percentages[2]
    # 验证是否没有其他缺失的值
print(avocado_data[['TotalBags', 'SmallBags', 'LargeBags', 'XLargeBags']].isnull().sum())

TotalBags     0
SmallBags     0
LargeBags     0
XLargeBags    0
dtype: int64

avocado_data[['plu4046', 'plu4225', 'plu4770']] = avocado_data[['plu4046', 'plu4225', 'plu4770']].astype(str)
avocado_data[['TotalBags', 'SmallBags', 'LargeBags', 'XLargeBags']] = avocado_data[['TotalBags', 'SmallBags', 'LargeBags', 'XLargeBags']].astype(int)

# 每个PLU代码的唯一值的数量
unique_counts = [avocado_data['plu4046'].nunique(), avocado_data['plu4225'].nunique(), avocado_data['plu4770'].nunique()]
plu_codes = ['PLU 4046', 'PLU 4225', 'PLU 4770']
barplot = sns.barplot(x=plu_codes, y=unique_counts, palette='viridis')
plt.title('Number of Unique Values for PLU Codes')
plt.xlabel('PLU Codes')
plt.ylabel('Number of Unique Values')
#在条形图的顶部添加数据值
for i, count in enumerate(unique_counts):
    barplot.text(i, count + 0.1, str(count), ha='center', va='bottom')
plt.show()

# 计算所有行中每种包类型的和
total_small_bags = avocado_data['SmallBags'].sum()
total_large_bags = avocado_data['LargeBags'].sum()
total_xlarge_bags = avocado_data['XLargeBags'].sum()
# 为饼图创建数据
sizes = [total_small_bags, total_large_bags, total_xlarge_bags]
labels = ['SmallBags', 'LargeBags', 'XLargeBags']
colors = ['lightcoral', 'lightskyblue', 'lightgreen']
# 绘制饼状图
plt.figure(figsize=(4, 4))
plt.pie(sizes, labels=labels, colors=colors, autopct='%1.1f%%', startangle=140)
plt.title('Distribution of Bag Types')
plt.show()

# 计算“type”的分布
type_distribution = avocado_data['type'].value_counts()
# 为饼状图创建数据
sizes = type_distribution.values
labels = type_distribution.index
colors = ['lightcoral', 'lightskyblue']
# 绘制饼状图
plt.figure(figsize=(4, 4))
plt.pie(sizes, labels=labels, colors=colors, autopct='%1.1f%%', startangle=140)
plt.title('Distribution of Avocado Types')
plt.show()

# 按“地区”分组，并计算每个地区的平均“平均价格”
avg_price_by_region = avocado_data.groupby('region')['AveragePrice'].mean()
sorted_avg_price_by_region = avg_price_by_region.sort_values(ascending=False)
top_regions = sorted_avg_price_by_region.head(10)
# 使用Plotly创建交互式水平条形图
fig = px.bar(x=top_regions.values, y=top_regions.index, orientation='h', color=top_regions.values,
labels={'x': 'Average Price', 'y': 'Region'}, title='Top Regions by Average Price',
color_continuous_scale='Plasma')
fig.show()

# 按“区域”分组，并计算每个区域的平均“TotalVolume”
avg_volume_by_region = avocado_data.groupby('region')['TotalVolume'].mean()
sorted_avg_volume_by_region = avg_volume_by_region.sort_values(ascending=False)
top_regions_volume = sorted_avg_volume_by_region.head(10)
# 使用Plotly创建交互式水平条形图
fig = px.bar(x=top_regions_volume.values, y=top_regions_volume.index, orientation='h', color=top_regions_volume.values,
labels={'x': 'Average Total Volume', 'y': 'Region'}, title='Top Regions by Average Total Volume',
color_continuous_scale='Jet')
fig.show()

# 过滤“TotalUS”区域的数据
total_us_data = avocado_data[avocado_data['region'] == 'TotalUS']
# 过滤其他区域的数据
other_regions_data = avocado_data[avocado_data['region'] != 'TotalUS']
# 计算“TotalUS”区域的“TotalVolume”之和
total_us_volume = total_us_data['TotalVolume'].sum()
# 计算其他区域的“TotalVolume”之和
other_regions_volume = other_regions_data.groupby('region')['TotalVolume'].sum().sum()
# 创建一个条形图来比较“TotalVolume”与“TotalUS”和其他
sns.barplot(x=['TotalUS', 'Other Regions'], y=[total_us_volume, other_regions_volume], palette='viridis')
plt.xlabel('Region')
plt.ylabel('Total Volume')
plt.title('Comparison of Total Volume for TotalUS and Other Regions')
plt.show()

# 可视化AveragePrice的分布
plt.figure(figsize=(6, 4))
sns.histplot(avocado_data['AveragePrice'], bins=30, kde=True, color='blue')
plt.title('Distribution of Average Price')
plt.show()

# 可视化TotalVolume的分布
plt.figure(figsize=(6, 4))
sns.histplot(avocado_data['TotalVolume'], bins=30, kde=True, color='green')
plt.title('Distribution of Total Volume')
plt.show()

# 将“日期”列转换为日期时间格式
avocado_data['Date'] = pd.to_datetime(avocado_data['Date'])
# 随时间变化的平均价格
plt.figure(figsize=(14, 6))
sns.lineplot(x='Date', y='AveragePrice', data=avocado_data, color='orange')
plt.title('Average Avocado Price Over Time')
plt.show()

# 交互式绘制平均价格随时间的变化
fig = px.line(avocado_data, x='Date', y='AveragePrice', title='Average Avocado Price Over Time', markers=True)
fig.update_layout(xaxis_title='Date', yaxis_title='Average Price')
fig.show()

# 绘制总体积随时间的变化
plt.figure(figsize=(14, 6))
sns.lineplot(x='Date', y='TotalVolume', data=avocado_data, color='purple')
plt.title('Total Avocado Volume Over Time')
plt.show()

# 交互式绘制平均价格随时间的变化
fig = px.line(avocado_data, x='Date', y='TotalVolume', title='Total Avocado Volume Over Time', markers=True)
fig.update_layout(xaxis_title='Date', yaxis_title='Total Volume')
fig.show()

# 设置自定义箱边
custom_bin_edges = [0, 50000, 100000, 150000, np.inf]
# 为TotalVolume创建箱子
avocado_data['VolumeCategory'] = pd.cut(avocado_data['TotalVolume'], bins=custom_bin_edges, labels=['Low', 'Medium', 'High', 'Very High'])
# 为“TotalBags”创建箱子
avocado_data['BagsCategory'] = pd.cut(avocado_data['TotalBags'], bins=custom_bin_edges, labels=['Low', 'Medium', 'High', 'Very High'])
fig, axes = plt.subplots(1, 2, figsize=(12, 6))
# “TotalVolume”的子图
ax1 = sns.countplot(x='VolumeCategory', data=avocado_data, palette='viridis', ax=axes[0])
ax1.set_title('Distribution of TotalVolume Categories')
ax1.set_xlabel('TotalVolume Category')
ax1.set_ylabel('Count')
# 为“TotalVolume”添加数据标签
for p in ax1.patches:
    ax1.annotate(f'{int(p.get_height())}', (p.get_x() + p.get_width() / 2., p.get_height()), ha='center', va='baseline', fontsize=10, color='black')
    # “TotalBags”的子图
    ax2 = sns.countplot(x='BagsCategory', data=avocado_data, palette='viridis', ax=axes[1])
    ax2.set_title('Distribution of TotalBags Categories')
    ax2.set_xlabel('TotalBags Category')
    ax2.set_ylabel('Count')
# 为“TotalBags”添加数据标签
for p in ax2.patches:
    ax2.annotate(f'{int(p.get_height())}', (p.get_x() + p.get_width() / 2., p.get_height()), ha='center', va='baseline', fontsize=10, color='black')
plt.tight_layout()
plt.show()

from sklearn.preprocessing import LabelEncoder
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_absolute_error
# 创建DataFrame的副本
X = avocado_data[['Date', 'type', 'region', 'VolumeCategory', 'BagsCategory']].copy()
y = avocado_data['AveragePrice']
# 提取年、月、日
X['Year'] = X['Date'].dt.year
X['Month'] = X['Date'].dt.month
X['Day'] = X['Date'].dt.day
# 删除原来的“Date”列
X = X.drop('Date', axis=1)
# 初始化LabelEncoder
label_encoder = LabelEncoder()
# 对分类列应用标签编码
X['type'] = label_encoder.fit_transform(X['type'])
X['region'] = label_encoder.fit_transform(X['region'])
X['VolumeCategory'] = label_encoder.fit_transform(X['VolumeCategory'])
X['BagsCategory'] = label_encoder.fit_transform(X['BagsCategory'])
# 将数据分成训练集和测试集
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
# 初始化和训练线性回归模型
model = LinearRegression()
model.fit(X_train, y_train)
# 对测试集进行预测
predictions = model.predict(X_test)
# 评估模型
mae = mean_absolute_error(y_test, predictions)
print(f'Mean Absolute Error: {mae}')

Mean Absolute Error: 0.21999210673812558

# 创建1行2列的子图
fig, axes = plt.subplots(nrows=1, ncols=2, figsize=(15, 6))
# 带图例的实际价格与预测价格的散点图
sns.scatterplot(x=y_test, y=predictions, label='Actual Prices', alpha=0.7, ax=axes[0])
sns.scatterplot(x=y_test, y=y_test, label='Predicted Prices', alpha=0.7, ax=axes[0])
axes[0].set_title('Actual vs. Predicted Prices with Legends')
axes[0].set_xlabel('Actual Prices')
axes[0].set_ylabel('Predicted Prices')
axes[0].legend()
# 带有图例的残差图
residuals = y_test - predictions
sns.scatterplot(x=predictions, y=residuals, label='Residuals', alpha=0.7, ax=axes[1])
axes[1].axhline(y=0, color='black', linestyle='--', label='Zero Residual Line')
axes[1].set_title('Residual Plot with Legends')
axes[1].set_xlabel('Predicted Prices')
axes[1].set_ylabel('Residuals')
axes[1].legend()
plt.tight_layout()
plt.show()

本次分析为牛油果数据集提供了有价值的见解，有助于更好地理解影响牛油果价格和数量的因素。 经过训练的线性回归模型在预测牛油果价格方面表现出了良好的性能，这一点可以通过评估指标和预测实际价值的可视化来证明。 进一步的调查、特征工程和对其他模型的探索可以增强预测能力和分析的深度。 这一分析为继续探索和改进模型奠定了基础，可能有助于更准确的预测和更深入地了解牛油果市场动态。