Deleting the point reduces the trend, but does not eliminate it.Alternatively, variable transformations are used to remove trends. This may improve the statistical properties of a regression model, but it complicates analysis and interpretation. Any transformation alters the economic meaning of a variable, favoring the predictive power of a model over explanatory simplicity.The manner of trend-removal depends on the type of trend. One type of trend is produced by a trend-stationary (TS) process, which is the sum of a deterministic trend and a stationary process. TS variables, once identified, are often linearized with a power or log transformation, then detrended by regressing on time. The detrend function, used above, removes the least-squares line from the data. This transformation often has the side effect of regularizing influential observations.