ARIMA预测后绘制柱状图遇数据类型不兼容问题求助

问题背景 #

使用ARIMA模型预测数据集未来5天的Amt数值，绘制包含现有数据与预测结果的柱状图时，触发类型不兼容错误。

示例数据集 #

CCY Pair        Time   Amt
0   GBPUSD  13/05/2023  1000
1   EURUSD  13/05/2023  2000
2   EURUSD  14/05/2023  3000
3   EURUSD  14/05/2023  5000
4   GBPEUR  15/05/2023  4000

报错详情 #

Traceback (most recent call last):
File "Graphs.py", line 46, in
plt.bar(combined_data.index, combined_data['Amt'])
File "/opt/homebrew/lib/python3.11/site-packages/matplotlib/pyplot.py", line 2439, in bar
return gca().bar(
^^^^^^^^^^
File "/opt/homebrew/lib/python3.11/site-packages/matplotlib/init.py", line 1442, in inner
return func(ax, *map(sanitize_sequence, args), **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/opt/homebrew/lib/python3.11/site-packages/matplotlib/axes/_axes.py", line 2460, in bar
raise TypeError(f'the dtypes of parameters x ({x.dtype}) '
TypeError: the dtypes of parameters x (object) and width (float64) are incompatible

原代码 #

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from statsmodels.tsa.arima.model import ARIMA


# The "Time" column contains the time periods, and "Amt" contains the values
df = pd.read_csv("Data.csv")

# Convert the "Time" column to a datetime type
df['Time'] = pd.to_datetime(df['Time'], dayfirst=True)

# Set the "Time" column as the index
df.set_index('Time', inplace=True)

# Sort the DataFrame by the index
df.sort_index(inplace=True)

df.index = pd.to_datetime(df.index).to_period('D')

# Prepare the data for ARIMA modeling
data = df['Amt']

# Fit the ARIMA model
model = ARIMA(data, order=(1,0,0)) 
model_fit = model.fit()

# Predict the next x periods
x = 5  # Number of periods to predict
predictions = model_fit.forecast(steps=x)

# Generate the next x bar charts based on the predictions
next_time_periods = pd.date_range(start=df.index.max().to_timestamp() + pd.DateOffset(days=1), periods=x, freq='D')  # Generate x future time periods
next_bar_charts = pd.DataFrame({'Time': next_time_periods, 'Amt': predictions}, index=next_time_periods)

# Concatenate current and predicted bar chart data
combined_data = pd.concat([df, next_bar_charts])

# Plot the combined bar chart
plt.figure(figsize=(10, 6))
plt.bar(combined_data.index, combined_data['Amt'])
plt.xlabel('Time')
plt.ylabel('Amt')
plt.xticks(rotation=45)  # Rotate x-axis labels for better readability
plt.show()

错误原因 #

索引类型不统一导致拼接后索引转为object类型，而matplotlib的plt.bar不支持该类型作为x轴参数：

• 原数据df的索引是Period类型（通过to_period('D')转换）
• 预测数据next_bar_charts的索引是Datetime类型
• 拼接后combined_data的索引自动转为object类型，触发类型不兼容报错。

解决方案 #

统一索引类型，确保原数据与预测数据的索引类型一致。以下提供两种可行方案：

方案1：统一为Datetime类型（推荐） #

注释掉原代码中转换为Period索引的行，保留Datetime索引：

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from statsmodels.tsa.arima.model import ARIMA


df = pd.read_csv("Data.csv")

# 转换Time列为datetime类型并设为索引
df['Time'] = pd.to_datetime(df['Time'], dayfirst=True)
df.set_index('Time', inplace=True)
df.sort_index(inplace=True)

# 注释掉转换为Period索引的代码
# df.index = pd.to_datetime(df.index).to_period('D')

# 准备ARIMA建模数据
data = df['Amt']

# 拟合ARIMA模型
model = ARIMA(data, order=(1,0,0)) 
model_fit = model.fit()

# 预测未来5天
x = 5
predictions = model_fit.forecast(steps=x)

# 生成未来时间周期（与原数据索引类型一致）
next_time_periods = pd.date_range(start=df.index.max() + pd.DateOffset(days=1), periods=x, freq='D')
next_bar_charts = pd.DataFrame({'Amt': predictions}, index=next_time_periods)

# 拼接数据
combined_data = pd.concat([df, next_bar_charts])

# 绘制柱状图
plt.figure(figsize=(10, 6))
plt.bar(combined_data.index, combined_data['Amt'])
plt.xlabel('时间')
plt.ylabel('金额')
plt.xticks(rotation=45)
plt.tight_layout()  # 防止标签被截断
plt.show()

方案2：统一为Period类型 #

将预测数据的索引转换为Period类型：

# 生成未来时间周期时转为Period类型
next_time_periods = pd.period_range(start=df.index.max() + pd.DateOffset(days=1), periods=x, freq='D')
next_bar_charts = pd.DataFrame({'Amt': predictions}, index=next_time_periods)

内容的提问来源于stack exchange，提问作者Tom Pitts