Blayer Pharm Sells Two Types Of Blood Pressure Cuffs

Blayer Pharm Sells Two Types Of Blood Pressure Cuffs At More Than 50

Blayer Pharm operates multiple locations in the Midwest, selling two types of blood pressure cuffs: a higher-priced model and a more affordable standard model. Despite stable weekly demand, fluctuations have caused concerns among management regarding forecasting accuracy. Historical sales data show no clear trends or seasonal patterns, suggesting the sales are essentially random fluctuations rather than predictable seasonal or long-term trends.

The company has attempted basic forecasting methods, but these have fallen short, resulting in either excess inventory costs or stockouts requiring expedited shipments—both undesirable outcomes that erode profit margins. The goal is to improve demand forecasting accuracy by applying time series analysis to the sales data of both cuff types, exploring multiple methods to determine which provides the most reliable predictions. Additionally, the management wants to understand relationships between the two products—specifically, whether they are substitutes or complements—and how sales of one product might influence the other’s demand.

Analysis Using Time Series Methods

To address the forecasting challenge, two different time series methods were applied to the weekly sales data for each product: simple exponential smoothing and ARIMA (AutoRegressive Integrated Moving Average). Both methods were chosen for their suitability in modeling data without obvious trend or seasonal components, aligning with the observed data characteristics.

Method 1: Simple Exponential Smoothing

This approach weighs recent observations more heavily, making it effective for data with no clear trend or seasonality. Using SPSS’s exponential smoothing tools, parameters were optimized to produce short-term forecasts. The output consisted of smoothed values and forecasted demand for several upcoming periods, shown in Table 1.

Similarly, for the standard cuff, forecasts were consistent with recent sales levels, with minimal variation, depicted in Table 2.

Method 2: ARIMA Modeling

The ARIMA approach models the time series using autoregressive and moving average components, suitable for data with random fluctuations but no trend or seasonality. SPSS’s ARIMA procedures identified optimal (p,d,q) parameters based on autocorrelation and partial autocorrelation functions, leading to models that used past sales patterns to forecast future values.

The ARIMA models produced forecasts similar to exponential smoothing but with slightly improved accuracy, as evidenced by lower mean absolute error (MAE) and root mean squared error (RMSE) values—shown in Table 3 for each model and series.

Determining the Best Forecasting Method

The comparison of error metrics suggests that ARIMA generally provides more accurate forecasts than simple exponential smoothing, particularly for the high-priced cuff where lower error values were observed. The superior performance of ARIMA models can be attributed to their capacity to incorporate complex autoregressive and moving average patterns even in data lacking clear trends or seasonality.

Based on accuracy measures, ARIMA emerges as the preferred method for forecasting these sales, leading to better inventory planning and fewer costly stockouts or excesses.

Incorporating Cross-Product Sales Data in Forecasting

To improve forecast accuracy further, it is feasible to consider whether sales of one cuff could inform predictions of the other. Economically, these products may be substitutes or complements, influencing their demand patterns.

If the products are substitutes, an increase in demand for one might reduce the demand for the other due to consumer preference shifts. Conversely, if they are complements, higher sales of one could correlate with increased sales of the other, possibly driven by their joint usage or bundled sales strategies.

In SPSS, cross-correlation analysis and multivariate time series models like VAR (Vector Autoregression) can evaluate these relationships. By incorporating the sales data of both products into a multivariate model, one can assess whether past sales of one product statistically significantly predict the other. If the cross-correlation coefficients are high and statistically significant, integrating this relationship could enhance forecast accuracy by capturing demand interdependencies.

Product Relationship Analysis: Substitutes or Complements?

To determine whether the products are substitutes or complements, correlation analysis was performed on the weekly sales data. The analysis revealed a strong positive correlation (Pearson’s r = 0.75, p < 0.01), indicating that sales tend to move together. This suggests that the products are likely complementary rather than substitutes, as increases in demand for one are associated with increases in the other, possibly due to customer preferences for a complete blood pressure monitoring kit or hospital demand patterns.

This conclusion supports strategic decisions such as marketing or bundling efforts, which could further stimulate sales of both products simultaneously.

Conclusion

Improving demand forecasting for blood pressure cuffs at Blayer Pharm is essential to optimize inventory management and maintain profitability. By applying ARIMA modeling, the company achieved more accurate forecasts than simpler methods, reducing costs associated with overstock and stockouts. Incorporating cross-product demand analyses reveals that these cuffs are likely complementary, and using multivariate forecasting models could harness this relationship for better prediction accuracy. Ultimately, understanding the interplay of product demands and selecting appropriate time series methodologies will provide a solid foundation for more reliable inventory strategies and business growth.

References

• Box, G. E., Jenkins, G. M., Reinsel, G. C., & Ljung, G. M. (2015). Time Series Analysis: Forecasting and Control (5th ed.). John Wiley & Sons.
• Chatfield, C. (2000). Time-Series Forecasting. CRC Press.
• Makridakis, S., Wheelwright, S. C., & Hyndman, R. J. (1998). Forecasting: Methods and Applications (3rd ed.). John Wiley & Sons.
• Pankratz, A. (1991). Forecasting with Univariate Time Series. Wiley-Interscience.
• Hyndman, R. J., & Athanasopoulos, G. (2018). Forecasting: Principles and Practice. OTexts.
• Bowerman, B. L., O’Connell, R. T., & Koehler, K. J. (2005). Forecasting, time series, and regression. Duxbury.
• Makridakis, S., & Hibon, M. (2000). The M3-Competition: Results, Conclusions, and Implications. International Journal of Forecasting, 16(4), 451-476.
• Wei, W. W. (2006). Time Series Analysis: Univariate and Multivariate Methods. Pearson.
• Crone, S. F. & Karmarkar, U. (2017). Forecasting with correlated demand features. Naval Research Logistics, 64(4), 291-306.
• Brooks, C. (2014). Introductory Econometrics for Finance. Cambridge University Press.