# Car rental time series forecasting [closed]

I have the time series of car rentals/demand in a location. The time axis is every hour for 3 months. The y-axis is number of car rentals/demand.

I want to do prediction for future hours given this information. I would like to know the prediction models for this type of data.

I am not sure about the performance of ARIMA or other time series predicion models given low values of car rentals at each hour (close to zero). The maximum is only 5 car rentals.

What are the typical and high accuracy methods? It could be statistical or machine-learning based models.

• If you have a trend (including a constant one) and cycles, exponential smoothing is worth a try. Jun 9, 2022 at 7:44
• Just looking at the graph, modelling as a Poisson random variable may be sufficient. 8 periods a day for ~90 days gives us 720 periods in the sample. So just guessing some Poisson values, we get dpois(0:5,0.8)*720 equals 323.5168542 258.8134833 103.5253933 27.6067716 5.5213543 0.8834167. The fit for predictions of 3/4/5 days are pretty close to your graph I believe by eye. Jun 9, 2022 at 11:12
• @Pcump_Ravenclaw : would you mind sharing your data ? Jun 9, 2022 at 12:09
• Are there zeros in your demand history? Do you anticipate needing any of the intermittent demand techniques? Jun 10, 2022 at 15:18
• @AndyW thanks for your reply. I don't entirely understand what you mean by " fit for predictions of 3/4/5 days are pretty close to your graph I believe by eye". I suppose that the numbers: "323.5168542 258.8134833 103.5253933 27.6067716 5.5213543 0.8834167" are the number of periods with 0,1,2,3,4& 5 cars over this entire 3 month time period? Jul 8, 2022 at 15:57