I'll produce predictions for US car sales by manufacture every month. There are already several blogs that describe the industry and sales that do a great job. Autoblog by the Numbers and Counting Cars are some to mention.
Unlike their analysis, I'll try to focus on the residuals (the stuff I can't predict) to tell the story. To highlight the difference, I think its instructive to look at what Autoblog mentioned.
The Autoblog article (link above) highlights Mitsubishi for increasing sales. However, my prediction for Mitsubishi sales are pretty much exactly what the sales were. In essence, given this model, we didn't learn much. On the other hand, Land Rover and Jaguar had the largest residuals (in percent terms) and Land Rover and Acura had the largest deviance (Residual / Variance). I think these results are more telling because we didn't predict them correctly; something might have changed.
I'll publish this on a new blog: datAutomotive.
Here is a Shiny App For Car Sales and below are graphs / tables of my current analysis and future predictions.
Predicted Values 9/14 | Actual Values 9/14 | log(Predicted/Actual) | Deviance | |
Acura | 488.036 | 576.333 | 0.166 | 1.731 |
Audi | 632.252 | 621.542 | -0.017 | 0.326 |
BMW | 1,061.329 | 1,066.083 | 0.004 | 0.032 |
Buick | 649.648 | 727.750 | 0.114 | 0.571 |
Cadillac | 568.021 | 576.208 | 0.014 | 0.089 |
Chevrolet | 6,284.594 | 6,411.375 | 0.020 | 0.135 |
Chrysler | 1,058.392 | 1,199.208 | 0.125 | 0.710 |
Dodge | 1,892.465 | 1,834.167 | -0.031 | 0.172 |
Ford | 7,374.669 | 7,177.542 | -0.027 | 0.285 |
GMC | 1,602.260 | 1,594.542 | -0.005 | 0.029 |
Honda | 4,434.780 | 4,349.625 | -0.019 | 0.174 |
Hyundai | 2,291.306 | 2,333.750 | 0.018 | 0.271 |
Infiniti | 315.217 | 326.542 | 0.035 | 0.234 |
Jaguar | 37.834 | 47.583 | 0.229 | 0.908 |
Jeep | 2,491.378 | 2,301.292 | -0.079 | 0.853 |
Kia | 1,853.734 | 1,692.833 | -0.091 | 1.115 |
Land.Rover | 159.975 | 129.417 | -0.212 | 1.816 |
Lexus | 1,011.967 | 910.500 | -0.106 | 1.189 |
Lincoln | 285.903 | 302.375 | 0.056 | 0.394 |
Mazda | 1,074.272 | 999.167 | -0.072 | 0.743 |
Mercedes.Benz | 1,186.220 | 1,230.125 | 0.036 | 0.364 |
Mini | 196.975 | 175.792 | -0.114 | 0.626 |
Mitsubishi | 231.582 | 231.583 | 0.00000 | 0.00002 |
Nissan | 4,176.995 | 3,963.250 | -0.053 | 0.516 |
Porsche | 156.496 | 150.292 | -0.040 | 0.408 |
Subaru | 1,727.597 | 1,729.875 | 0.001 | 0.026 |
Toyota | 6,928.472 | 6,059.458 | -0.134 | 1.309 |
Volkswagen | 1,123.412 | 1,083.167 | -0.036 | 0.407 |
Volvo | 161.531 | 194.458 | 0.186 | 1.070 |
Predicted Values for 10/14 | |
Acura | 521.878 |
Audi | 590.125 |
BMW | 1,051.492 |
Buick | 666.165 |
Cadillac | 533.468 |
Chevrolet | 5,693.145 |
Chrysler | 1,097.288 |
Dodge | 1,493.426 |
Ford | 6,694.012 |
GMC | 1,572.525 |
Honda | 4,049.582 |
Hyundai | 1,998.896 |
Infiniti | 262.475 |
Jaguar | 40.422 |
Jeep | 2,132.652 |
Kia | 1,650.236 |
Land.Rover | 160.884 |
Lexus | 944.946 |
Lincoln | 272.964 |
Mazda | 894.234 |
Mercedes.Benz | 1,184.991 |
Mini | 183.743 |
Mitsubishi | 193.982 |
Nissan | 3,629.955 |
Porsche | 158.540 |
Subaru | 1,694.661 |
Toyota | 5,895.316 |
Volkswagen | 923.242 |
Volvo | 153.691 |
Predicting such things as car sales, either in aggregate or by make/model, based on time series data is a failed effort. Car sales, in aggregate, are a function of median income and interest rate. At the make/model level, they're a function of specific events, e.g. infotainment system, GPS, ABS, overall design. Time series tells one nothing about future events which will drive increased sales of specific makes/models.
ReplyDeleteI think you're wrong. You can learn somethings from past observations to learn about the future (foremost: trends and seasonality). And just about all time series analysis takes into account previous values.
DeleteI agree there's a lot the model isn't capturing (and there for sure is room for improvement), but to suggest that time series can't predict ANYTHING is bizarre.
For example look at the Audi Time Series here: https://sweiss.shinyapps.io/CarSales/.
This model takes into account seasonal and growth factors that have been fairly stable over time. It seems natural (as a first approximation) to assume these trends will continue.