Forecasting medicine sales using an ARIMA model

Abstract

This study aims to apply the ARIMA (AutoRegressive Integrated Moving Average) model to predict the drug sales at Chulabhorn Hospital using monthly sales data of 200 consistently selling drugs from August 1, 2019, to August 31, 2024. The goal is to create a model that can accurately forecast future sales. The model's performance is evaluated using the MAPE (Mean Absolute Percentage Error) metric to measure prediction accuracy.

The research process begins by loading the monthly sales data, followed by testing for stationarity using the Augmented Dickey-Fuller (ADF) test. If the data is non-stationary, differencing is applied to correct it. The data is then split into a training set (80%) and a test set (20%). The auto ARIMA function is used to automatically build the model and select the most appropriate parameters. The model is evaluated using MAE, RMSE, and MAPE metrics.

The research results show that the average MAPE for the ARIMA model is 19.6%, with a maximum error of 33.80% and a minimum error of 6.13%. These results indicate that the ARIMA model provides an acceptable level of accuracy in predicting drug sales. However, some periods exhibit higher-than-normal error rates.

The recommendation from this study is to develop a hybrid model combining ARIMA and Neural Networks to improve accuracy by capturing both linear and non-linear trends simultaneously.

References

Box, G. E., Jenkins, G. M., & Reinsel, G. C. (2015). Time Series Analysis: Forecasting and Control. John Wiley & Sons.

Hyndman, R. J., & Athanasopoulos, G. (2018). Forecasting: Principles and Practice. OTexts.

Rathipriya, R., Rahman, A. A. A., Dhamodharavadhani, S., Meero, A., & Yoganandan, G. (2022). "Demand forecasting model for time-series pharmaceutical data using shallow and deep neural network model," Neural Computing and Applications, Springer-Verlag London Ltd. DOI: 10.1007/s00521-022-07889- 9(521_2022_Article_7889).

Vongspanich, W. (2021). "Forecasting Medical Time Series Data Using the ARIMA Model," Siriraj Medical Bulletin, vol. 14, no. 4, pp. 38-49.

Rodríguez González, R., Abdul Rahman, A. A., & Aziz, A. (2021). Forecast of the demand for med ications by a pharmaceutical company. Procedia Computer Science, 179, 480-487. https://doi. org/10.1016/j.procs.2021.01.036

Ankita M., Srinivas, A., Soni, A., Prajapati, G., & Manjunath, P. S. (2024). "Pharmaceutical Sales Data Prediction Using Time Series Forecasting," Interna tional Journal of Intelligent Systems and Applications in Engineering, 12(13s), 681-696.

Dave, E., Leonardo, A., Jeanice, M., & Hanafiah, N. (2021). "Forecasting Indonesia Exports using a Hybrid Model ARIMA-LSTM," Procedia Computer Science, 179, 480-487