1) Why is differencing needed after applying a logarithmic transform? The incorrect transformation is used. Transformation methods only reduces non-stationary patterns. The logarithmic transform of a time series always has a seasonality component. 2) The output of the adfuller function for the data frame after differencing the log transformed data is given below. Does sufficient evidence exist to conclude that the time series is stationary at the a = 0.1 significance level? (-2.717130598388165, 0.07112054815085335, 14, 128, {'1%': -3.4825006939887997, '10%': -2.578960197753906, '5%': -2.884397984161377}, -440.35846985568105) Yes No 3) Can the residual r^t can be modeled directly? Yes No