Consider the data given in adoptionseries2_with_noise.xlsx. N(t) is in millions. So if N(t) = 0.05 that means there were 50000 adopters in period "t". These data correspond to the data reported in the second column (the "shuffled" data) on Assessing Primary Demand Slide 12. As we discussed in class, if the adoption series is noisy, the Bass Model estimates can be incorrect, which leads to serious forecasting error as depicted on Slide 13. The incorrectness is more with the "M" parameter rather than the "p" and "q" parameters, but because the three terms are interdependent, the incorrectness in the "M" parameter has some impact also on the "p" and "q" parameters. As a workaround, we can determine M from survey research or some other outside data, fix M to the value that was determined, and use the adoption series dataset to estimate just the "p" and "q". This requires the use of nonlinear regression as we saw in class