在 R 中使用 Prophet 进行预测
Forecasting using Prophet in R
我有这个包含 78 个不同位置的数据集，我必须在其中预测未来 3 个月单个产品的价格。数据有大约 6 个月（4 月至 10 月）的每日条目。但是，当我 运行 模型时，它给了我一个错误。请帮我解决这个错误。
或者有没有其他方法可以预测数据？
我的代码：
Area_forecast <- data1 %>%
filter(product == "garri") %>%
select( level2code, ds, y = pred) %>%
mutate(ds = paste(ds, "01", "01", sep = "-") %>% as.Date()) %>%
group_by(level2code) %>%
do(model = prophet(.),
Predicted_values = make_future_dataframe(model,92),
results = predict(model, Predicted_values))
错误：
Disabling yearly seasonality. Run prophet with yearly.seasonality=TRUE to override this.
Disabling daily seasonality. Run prophet
with daily.seasonality=TRUE to override this.
Error: object of type 'closure' is not subsettable
Sample data set: Data is divided into 78 areas with daily values.
ds product level2code pred
2021-04-01 garri NG020001 366.1540473
2021-04-01 garri NG020002 393.3557317
2021-04-01 garri NG020003 374.3132972
2021-04-01 garri NG021001 353.1266783
2021-04-01 garri NG021002 329.9673004
2021-04-01 garri NG021003 391.1370065
2021-04-01 garri NG021004 306.1356683
2021-04-01 garri NG020004 372.8115048
2021-04-01 garri NG020005 350.0100309
2021-04-01 garri NG021005 373.0025842
2021-04-01 garri NG020006 383.1839042
2021-04-01 garri NG021006 383.5212653
2021-04-01 garri NG020007 486.8226379
2021-04-01 garri NG020008 334.8151487
2021-04-01 garri NG021007 379.89125
2021-04-01 garri NG021008 370.9376296
2021-04-01 garri NG021009 389.4640031
2021-04-01 garri NG021010 365.9266992
2021-04-01 garri NG020009 380.4655716
2021-04-01 garri NG020010 352.6139069
2021-04-01 garri NG020011 243.5701952
2021-04-01 garri NG021011 366.6883205
2021-04-01 garri NG021012 326.9402671
2021-04-01 garri NG020012 403.5666062
2021-04-01 garri NG021013 405.2860516
2021-04-01 garri NG021014 348.023662
2021-04-01 garri NG020013 372.7586167
2021-04-01 garri NG020014 392.5432373
2021-04-01 garri NG020015 362.1098193
2021-04-01 garri NG020016 343.9205353
2021-04-01 garri NG020017 366.5290601
2021-04-01 garri NG020018 431.0832709
2021-04-01 garri NG020019 411.3581413
2021-04-01 garri NG021015 348.3274897
2021-04-01 garri NG021016 336.3743799
2021-04-01 garri NG020020 393.8901327
2021-04-01 garri NG021017 391.7393648
2021-04-01 garri NG021018 346.0988762
2021-04-01 garri NG021019 346.1310667
2021-04-01 garri NG021020 359.2837766
2021-04-01 garri NG020021 450.4413674
2021-04-01 garri NG020022 373.6406838
2021-04-01 garri NG021021 420.0838824
2021-04-01 garri NG020023 408.6912783
2021-04-01 garri NG020024 285.8249281
2021-04-01 garri NG020025 388.3307999
2021-04-01 garri NG020026 365.1686898
2021-04-01 garri NG020027 391.6867595
2021-04-01 garri NG021022 384.1850593
2021-04-01 garri NG021023 397.7172044
2021-04-01 garri NG020028 331.2182171
2021-04-01 garri NG021024 393.323333
2021-04-01 garri NG020029 405.5813127
2021-04-01 garri NG021025 359.658715
2021-04-01 garri NG021026 370.3166839
2021-04-01 garri NG021027 391.088788
2021-04-01 garri NG021028 364.2770249
2021-04-01 garri NG020030 375.2600611
2021-04-01 garri NG021029 358.9618615
2021-04-01 garri NG020031 439.1322423
2021-04-01 garri NG020032 402.7340519
2021-04-01 garri NG021030 345.2253201
2021-04-01 garri NG020033 356.2798309
2021-04-01 garri NG020034 369.3260162
2021-04-01 garri NG021031 368.6820545
2021-04-01 garri NG021032 342.8985896
2021-04-01 garri NG021033 388.3937476
2021-04-01 garri NG020035 369.5370586
2021-04-01 garri NG020036 374.2149029
2021-04-01 garri NG020037 374.3105323
2021-04-01 garri NG020038 419.9881972
2021-04-01 garri NG020039 356.6891081
2021-04-01 garri NG020040 385.9381759
2021-04-01 garri NG020041 308.1801669
2021-04-01 garri NG020042 383.1544687
2021-04-01 garri NG020043 381.4891835
2021-04-01 garri NG020044 362.6149694
2021-04-01 garri NG021034 351.0315915
2021-04-02 garri NG020001 347.5080909
2021-04-02 garri NG020002 417.7950263
2021-04-02 garri NG020003 357.6293544
2021-04-02 garri NG021001 364.0809209
2021-04-02 garri NG021002 350.5885553
2021-04-02 garri NG021003 412.9410955
2021-04-02 garri NG021004 326.3732426
2021-04-02 garri NG020004 374.9451842
2021-04-02 garri NG020005 335.0886535
2021-04-02 garri NG021005 365.002555
2021-04-02 garri NG020006 392.6334097
2021-04-02 garri NG021006 379.2592638
2021-04-02 garri NG020007 438.0161714
2021-04-02 garri NG020008 361.915949
2021-04-02 garri NG021007 357.060549
2021-04-02 garri NG021008 382.9647718
2021-04-02 garri NG021009 389.1019382
2021-04-02 garri NG021010 333.3285699
2021-04-02 garri NG020009 406.0860302
2021-04-02 garri NG020010 342.7835896
2021-04-02 garri NG020011 243.602651
2021-04-02 garri NG021011 358.4753587
2021-04-02 garri NG021012 333.9547987
2021-04-02 garri NG020012 424.8915992
2021-04-02 garri NG021013 407.2047254
2021-04-02 garri NG021014 356.6045034
2021-04-02 garri NG020013 358.297996
2021-04-02 garri NG020014 432.0024885
2021-04-02 garri NG020015 363.1590239
2021-04-02 garri NG020016 334.1112816
2021-04-02 garri NG020017 376.0309518
2021-04-02 garri NG020018 426.4144669
2021-04-02 garri NG020019 400.1660285
2021-04-02 garri NG021015 337.3337231
2021-04-02 garri NG021016 341.8770212
2021-04-02 garri NG020020 391.1870559
2021-04-02 garri NG021017 397.611482
2021-04-02 garri NG021018 345.1756482
2021-04-02 garri NG021019 346.5591129
2021-04-02 garri NG021020 347.5115345
2021-04-02 garri NG020021 435.024338
2021-04-02 garri NG020022 382.7996258
2021-04-02 garri NG021021 388.406993
2021-04-02 garri NG020023 417.0865963
2021-04-02 garri NG020024 293.7273414
2021-04-02 garri NG020025 408.075575
2021-04-02 garri NG020026 360.3596338
2021-04-02 garri NG020027 398.2523865
2021-04-02 garri NG021022 395.1357617
2021-04-02 garri NG021023 387.7742073
2021-04-02 garri NG020028 341.8796137
2021-04-02 garri NG021024 383.3824892
2021-04-02 garri NG020029 415.2975719
2021-04-02 garri NG021025 367.5180956
2021-04-02 garri NG021026 373.8995221
2021-04-02 garri NG021027 390.6734957
2021-04-02 garri NG021028 348.9460594
2021-04-02 garri NG020030 397.2809552
2021-04-02 garri NG021029 352.6791518
2021-04-02 garri NG020031 431.4367737
2021-04-02 garri NG020032 394.5424715
2021-04-02 garri NG021030 388.4691285
2021-04-02 garri NG020033 350.0768082
2021-04-02 garri NG020034 370.2760778
2021-04-02 garri NG021031 365.5463449
2021-04-02 garri NG021032 353.6569909
2021-04-02 garri NG021033 347.2475542
2021-04-02 garri NG020035 371.0247191
2021-04-02 garri NG020036 380.1494315
2021-04-02 garri NG020037 395.302709
2021-04-02 garri NG020038 433.3253765
2021-04-02 garri NG020039 350.8312732
2021-04-02 garri NG020040 382.1718306
2021-04-02 garri NG020041 316.6798756
2021-04-02 garri NG020042 409.7501652
2021-04-02 garri NG020043 385.8142597
2021-04-02 garri NG020044 379.9411943
2021-04-02 garri NG021034 345.6267946
而不是 do，使用 nest_by 并确保突变的结果被捕获在列表中。然后你可以一次搞定所有事情。
这基于您的示例数据。如果您有大量数据和 2 级代码，请去喝杯咖啡或吃午餐，因为这需要一段时间。只是这个例子在我的机器上花了几分钟。
Area_forecast <- data1 %>%
filter(product == "garri") %>%
select( level2code, ds, y = pred) %>%
mutate(ds = paste(ds, "01", "01", sep = "-") %>% as.Date()) %>%
nest_by(level2code) %>%
mutate(mod = list(prophet(data)),
Predicted_values = list(make_future_dataframe(mod,92)),
results = list(predict(mod, Predicted_values)))
Area_forecast
# A tibble: 78 x 5
# Rowwise: level2code
level2code data mod Predicted_values results
<chr> <list<tibble[,2]>> <list> <list> <list>
1 NG020001 [2 x 2] <prophet [31]> <df [12 x 1]> <df [12 x 16]>
2 NG020002 [2 x 2] <prophet [31]> <df [12 x 1]> <df [12 x 16]>
3 NG020003 [2 x 2] <prophet [31]> <df [12 x 1]> <df [12 x 16]>
4 NG020004 [2 x 2] <prophet [31]> <df [12 x 1]> <df [12 x 16]>
5 NG020005 [2 x 2] <prophet [31]> <df [12 x 1]> <df [12 x 16]>
6 NG020006 [2 x 2] <prophet [31]> <df [12 x 1]> <df [12 x 16]>
7 NG020007 [2 x 2] <prophet [31]> <df [12 x 1]> <df [12 x 16]>
8 NG020008 [2 x 2] <prophet [31]> <df [12 x 1]> <df [12 x 16]>
9 NG020009 [2 x 2] <prophet [31]> <df [12 x 1]> <df [12 x 16]>
10 NG020010 [2 x 2] <prophet [31]> <df [12 x 1]> <df [12 x 16]>
# ... with 68 more rows
如果您想查看 level2code 的个别结果，您可以使用：
Area_forecast %>%
select(level2code, results) %>%
filter(level2code == "NG020001") %>%
unnest(results)
# A tibble: 12 x 17
# Groups: level2code [1]
level2code ds trend zeros zeros_lower zeros_upper additive_terms additive_terms_lower
<chr> <dttm> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 NG020001 2021-04-01 00:00:00 366. 0 0 0 0 0
2 NG020001 2021-04-02 00:00:00 348. 0 0 0 0 0
3 NG020001 2021-04-03 00:00:00 329. 0 0 0 0 0
4 NG020001 2021-04-04 00:00:00 310. 0 0 0 0 0
5 NG020001 2021-04-05 00:00:00 292. 0 0 0 0 0
6 NG020001 2021-04-06 00:00:00 273. 0 0 0 0 0
7 NG020001 2021-04-07 00:00:00 254. 0 0 0 0 0
8 NG020001 2021-04-08 00:00:00 236. 0 0 0 0 0
9 NG020001 2021-04-09 00:00:00 217. 0 0 0 0 0
10 NG020001 2021-04-10 00:00:00 198. 0 0 0 0 0
11 NG020001 2021-04-11 00:00:00 180. 0 0 0 0 0
12 NG020001 2021-04-12 00:00:00 161. 0 0 0 0 0
# ... with 9 more variables: additive_terms_upper <dbl>, multiplicative_terms <dbl>,
# multiplicative_terms_lower <dbl>, multiplicative_terms_upper <dbl>, yhat_lower <dbl>, yhat_upper <dbl>,
# trend_lower <dbl>, trend_upper <dbl>, yhat <dbl>
我有这个包含 78 个不同位置的数据集，我必须在其中预测未来 3 个月单个产品的价格。数据有大约 6 个月（4 月至 10 月）的每日条目。但是，当我 运行 模型时，它给了我一个错误。请帮我解决这个错误。
或者有没有其他方法可以预测数据？ 我的代码：
Area_forecast <- data1 %>%
filter(product == "garri") %>%
select( level2code, ds, y = pred) %>%
mutate(ds = paste(ds, "01", "01", sep = "-") %>% as.Date()) %>%
group_by(level2code) %>%
do(model = prophet(.),
Predicted_values = make_future_dataframe(model,92),
results = predict(model, Predicted_values))
错误：
Disabling yearly seasonality. Run prophet with yearly.seasonality=TRUE to override this.
Disabling daily seasonality. Run prophet
with daily.seasonality=TRUE to override this.
Error: object of type 'closure' is not subsettable
Sample data set: Data is divided into 78 areas with daily values.
ds product level2code pred
2021-04-01 garri NG020001 366.1540473
2021-04-01 garri NG020002 393.3557317
2021-04-01 garri NG020003 374.3132972
2021-04-01 garri NG021001 353.1266783
2021-04-01 garri NG021002 329.9673004
2021-04-01 garri NG021003 391.1370065
2021-04-01 garri NG021004 306.1356683
2021-04-01 garri NG020004 372.8115048
2021-04-01 garri NG020005 350.0100309
2021-04-01 garri NG021005 373.0025842
2021-04-01 garri NG020006 383.1839042
2021-04-01 garri NG021006 383.5212653
2021-04-01 garri NG020007 486.8226379
2021-04-01 garri NG020008 334.8151487
2021-04-01 garri NG021007 379.89125
2021-04-01 garri NG021008 370.9376296
2021-04-01 garri NG021009 389.4640031
2021-04-01 garri NG021010 365.9266992
2021-04-01 garri NG020009 380.4655716
2021-04-01 garri NG020010 352.6139069
2021-04-01 garri NG020011 243.5701952
2021-04-01 garri NG021011 366.6883205
2021-04-01 garri NG021012 326.9402671
2021-04-01 garri NG020012 403.5666062
2021-04-01 garri NG021013 405.2860516
2021-04-01 garri NG021014 348.023662
2021-04-01 garri NG020013 372.7586167
2021-04-01 garri NG020014 392.5432373
2021-04-01 garri NG020015 362.1098193
2021-04-01 garri NG020016 343.9205353
2021-04-01 garri NG020017 366.5290601
2021-04-01 garri NG020018 431.0832709
2021-04-01 garri NG020019 411.3581413
2021-04-01 garri NG021015 348.3274897
2021-04-01 garri NG021016 336.3743799
2021-04-01 garri NG020020 393.8901327
2021-04-01 garri NG021017 391.7393648
2021-04-01 garri NG021018 346.0988762
2021-04-01 garri NG021019 346.1310667
2021-04-01 garri NG021020 359.2837766
2021-04-01 garri NG020021 450.4413674
2021-04-01 garri NG020022 373.6406838
2021-04-01 garri NG021021 420.0838824
2021-04-01 garri NG020023 408.6912783
2021-04-01 garri NG020024 285.8249281
2021-04-01 garri NG020025 388.3307999
2021-04-01 garri NG020026 365.1686898
2021-04-01 garri NG020027 391.6867595
2021-04-01 garri NG021022 384.1850593
2021-04-01 garri NG021023 397.7172044
2021-04-01 garri NG020028 331.2182171
2021-04-01 garri NG021024 393.323333
2021-04-01 garri NG020029 405.5813127
2021-04-01 garri NG021025 359.658715
2021-04-01 garri NG021026 370.3166839
2021-04-01 garri NG021027 391.088788
2021-04-01 garri NG021028 364.2770249
2021-04-01 garri NG020030 375.2600611
2021-04-01 garri NG021029 358.9618615
2021-04-01 garri NG020031 439.1322423
2021-04-01 garri NG020032 402.7340519
2021-04-01 garri NG021030 345.2253201
2021-04-01 garri NG020033 356.2798309
2021-04-01 garri NG020034 369.3260162
2021-04-01 garri NG021031 368.6820545
2021-04-01 garri NG021032 342.8985896
2021-04-01 garri NG021033 388.3937476
2021-04-01 garri NG020035 369.5370586
2021-04-01 garri NG020036 374.2149029
2021-04-01 garri NG020037 374.3105323
2021-04-01 garri NG020038 419.9881972
2021-04-01 garri NG020039 356.6891081
2021-04-01 garri NG020040 385.9381759
2021-04-01 garri NG020041 308.1801669
2021-04-01 garri NG020042 383.1544687
2021-04-01 garri NG020043 381.4891835
2021-04-01 garri NG020044 362.6149694
2021-04-01 garri NG021034 351.0315915
2021-04-02 garri NG020001 347.5080909
2021-04-02 garri NG020002 417.7950263
2021-04-02 garri NG020003 357.6293544
2021-04-02 garri NG021001 364.0809209
2021-04-02 garri NG021002 350.5885553
2021-04-02 garri NG021003 412.9410955
2021-04-02 garri NG021004 326.3732426
2021-04-02 garri NG020004 374.9451842
2021-04-02 garri NG020005 335.0886535
2021-04-02 garri NG021005 365.002555
2021-04-02 garri NG020006 392.6334097
2021-04-02 garri NG021006 379.2592638
2021-04-02 garri NG020007 438.0161714
2021-04-02 garri NG020008 361.915949
2021-04-02 garri NG021007 357.060549
2021-04-02 garri NG021008 382.9647718
2021-04-02 garri NG021009 389.1019382
2021-04-02 garri NG021010 333.3285699
2021-04-02 garri NG020009 406.0860302
2021-04-02 garri NG020010 342.7835896
2021-04-02 garri NG020011 243.602651
2021-04-02 garri NG021011 358.4753587
2021-04-02 garri NG021012 333.9547987
2021-04-02 garri NG020012 424.8915992
2021-04-02 garri NG021013 407.2047254
2021-04-02 garri NG021014 356.6045034
2021-04-02 garri NG020013 358.297996
2021-04-02 garri NG020014 432.0024885
2021-04-02 garri NG020015 363.1590239
2021-04-02 garri NG020016 334.1112816
2021-04-02 garri NG020017 376.0309518
2021-04-02 garri NG020018 426.4144669
2021-04-02 garri NG020019 400.1660285
2021-04-02 garri NG021015 337.3337231
2021-04-02 garri NG021016 341.8770212
2021-04-02 garri NG020020 391.1870559
2021-04-02 garri NG021017 397.611482
2021-04-02 garri NG021018 345.1756482
2021-04-02 garri NG021019 346.5591129
2021-04-02 garri NG021020 347.5115345
2021-04-02 garri NG020021 435.024338
2021-04-02 garri NG020022 382.7996258
2021-04-02 garri NG021021 388.406993
2021-04-02 garri NG020023 417.0865963
2021-04-02 garri NG020024 293.7273414
2021-04-02 garri NG020025 408.075575
2021-04-02 garri NG020026 360.3596338
2021-04-02 garri NG020027 398.2523865
2021-04-02 garri NG021022 395.1357617
2021-04-02 garri NG021023 387.7742073
2021-04-02 garri NG020028 341.8796137
2021-04-02 garri NG021024 383.3824892
2021-04-02 garri NG020029 415.2975719
2021-04-02 garri NG021025 367.5180956
2021-04-02 garri NG021026 373.8995221
2021-04-02 garri NG021027 390.6734957
2021-04-02 garri NG021028 348.9460594
2021-04-02 garri NG020030 397.2809552
2021-04-02 garri NG021029 352.6791518
2021-04-02 garri NG020031 431.4367737
2021-04-02 garri NG020032 394.5424715
2021-04-02 garri NG021030 388.4691285
2021-04-02 garri NG020033 350.0768082
2021-04-02 garri NG020034 370.2760778
2021-04-02 garri NG021031 365.5463449
2021-04-02 garri NG021032 353.6569909
2021-04-02 garri NG021033 347.2475542
2021-04-02 garri NG020035 371.0247191
2021-04-02 garri NG020036 380.1494315
2021-04-02 garri NG020037 395.302709
2021-04-02 garri NG020038 433.3253765
2021-04-02 garri NG020039 350.8312732
2021-04-02 garri NG020040 382.1718306
2021-04-02 garri NG020041 316.6798756
2021-04-02 garri NG020042 409.7501652
2021-04-02 garri NG020043 385.8142597
2021-04-02 garri NG020044 379.9411943
2021-04-02 garri NG021034 345.6267946
而不是 do，使用 nest_by 并确保突变的结果被捕获在列表中。然后你可以一次搞定所有事情。
这基于您的示例数据。如果您有大量数据和 2 级代码，请去喝杯咖啡或吃午餐，因为这需要一段时间。只是这个例子在我的机器上花了几分钟。
Area_forecast <- data1 %>%
filter(product == "garri") %>%
select( level2code, ds, y = pred) %>%
mutate(ds = paste(ds, "01", "01", sep = "-") %>% as.Date()) %>%
nest_by(level2code) %>%
mutate(mod = list(prophet(data)),
Predicted_values = list(make_future_dataframe(mod,92)),
results = list(predict(mod, Predicted_values)))
Area_forecast
# A tibble: 78 x 5
# Rowwise: level2code
level2code data mod Predicted_values results
<chr> <list<tibble[,2]>> <list> <list> <list>
1 NG020001 [2 x 2] <prophet [31]> <df [12 x 1]> <df [12 x 16]>
2 NG020002 [2 x 2] <prophet [31]> <df [12 x 1]> <df [12 x 16]>
3 NG020003 [2 x 2] <prophet [31]> <df [12 x 1]> <df [12 x 16]>
4 NG020004 [2 x 2] <prophet [31]> <df [12 x 1]> <df [12 x 16]>
5 NG020005 [2 x 2] <prophet [31]> <df [12 x 1]> <df [12 x 16]>
6 NG020006 [2 x 2] <prophet [31]> <df [12 x 1]> <df [12 x 16]>
7 NG020007 [2 x 2] <prophet [31]> <df [12 x 1]> <df [12 x 16]>
8 NG020008 [2 x 2] <prophet [31]> <df [12 x 1]> <df [12 x 16]>
9 NG020009 [2 x 2] <prophet [31]> <df [12 x 1]> <df [12 x 16]>
10 NG020010 [2 x 2] <prophet [31]> <df [12 x 1]> <df [12 x 16]>
# ... with 68 more rows
如果您想查看 level2code 的个别结果，您可以使用：
Area_forecast %>%
select(level2code, results) %>%
filter(level2code == "NG020001") %>%
unnest(results)
# A tibble: 12 x 17
# Groups: level2code [1]
level2code ds trend zeros zeros_lower zeros_upper additive_terms additive_terms_lower
<chr> <dttm> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 NG020001 2021-04-01 00:00:00 366. 0 0 0 0 0
2 NG020001 2021-04-02 00:00:00 348. 0 0 0 0 0
3 NG020001 2021-04-03 00:00:00 329. 0 0 0 0 0
4 NG020001 2021-04-04 00:00:00 310. 0 0 0 0 0
5 NG020001 2021-04-05 00:00:00 292. 0 0 0 0 0
6 NG020001 2021-04-06 00:00:00 273. 0 0 0 0 0
7 NG020001 2021-04-07 00:00:00 254. 0 0 0 0 0
8 NG020001 2021-04-08 00:00:00 236. 0 0 0 0 0
9 NG020001 2021-04-09 00:00:00 217. 0 0 0 0 0
10 NG020001 2021-04-10 00:00:00 198. 0 0 0 0 0
11 NG020001 2021-04-11 00:00:00 180. 0 0 0 0 0
12 NG020001 2021-04-12 00:00:00 161. 0 0 0 0 0
# ... with 9 more variables: additive_terms_upper <dbl>, multiplicative_terms <dbl>,
# multiplicative_terms_lower <dbl>, multiplicative_terms_upper <dbl>, yhat_lower <dbl>, yhat_upper <dbl>,
# trend_lower <dbl>, trend_upper <dbl>, yhat <dbl>