Diffusion Model for Cellular Phones and Hybrid Cars

MB 555

Problem Set Questions for Bass Diffusion Model

Problem 1 – Annual Cellular Telephone Subscribers 1985-2008

Although the technology was introduced in 1983, the data set starts in 1985. Use the regression approach to model the diffusion process and then answer the following questions. Turn in your model and your answers.

1. Model the diffusion curve

2. Show the graph of predicted versus actual

3. What are the MAPE and RMSE for your model?

MAPE = 918,238.36

RMSE = 99%

4. Do you get a better fit if you shift the predicted adopters down one period (down one row so the predicted are predicting the “next period”? Why?

RMSE = 736,579.87

MAPE = 61%

After predicted data was shifted, forecast became close to actual data and errors between them decreased. It could be because technology was introduced in 1983 and data set that we used starts in 1985. Thus, there would be higher number of actual subscribers and data would shift.

5. What do p & q tell you about the diffusion process of this technology?

P = 0.00448

q = 0.2689

Evaluations are based on industry average value of p=0.003 and q=0.38.

Based on these results it is expected that potential customers will become interested in the product through word of the mouth more effectively then through media advertisement. Also, market has reached its maturity and is dropping since p and q values are lower than industry average value.

6. The actual data show the first apparent peak in non cumulative was in 99-00 and then a second peak around 2004-05. What’s your best guess about the reasons for the second peak?

Introduction of web ready phones 1999 and overall drop in technology prices attracted significant number of customers that created a spike in sales.

(Sales of Internet-ready Cellular phones at retail locations have managed to take off at an astounding rate in 1999. Almost half (48 percent) of cell phones purchased at retail locations during the...