I analyze the development of the total number of users over time. It is obvious that a linear regression fits the data. The r squared is then also 0.9961.

Now the problem: a plot of the residuals shows the presence of a pattern. This should not be the case if the model would be correct.

Does anyone have experience how to solve this issue? I tried time series model fit, which gave me a random walk model (ARIMA 0,1,0) but I am not able to find out the coefficients of this model, which is exactly what I need.

Thanks.