Trend Line allow the user to identify the trends on data. Trend lines also called as "Best Fit Lines" as in this we will compute the lines that identify the trends.
Tableau uses 5 different trend models to compute trend lines
Linear is a default trend model in tableau. Linear relationship is used to describe straight relation between two variables. Mathematically, Linear relation is that satisfies the equation:
y = mx + b
Graphical representation Drag order date in column shelf and quantity in rows . click on Analytic tab > trend line
it will show straight line.
In logarithmic, Tableau will use either Natural logarithmic or base logarithmic to compute the trend line
Mathematical Expression use for that is
y = a + b * Ln (x) ------ natural logarithmic
y = a +b * log to the base 10(x) ---- base logarithmic
Drag ship date in column and quantity in row shelf then click Analytic tab > trendlines > choose the trend model
While using Logarithmic trend , First we need to filter out negative values as this model doesn't support negative values.
Trends lines execute at tableau level.
With Exponential model ,variable is transformed by natural log before estimation of model to compute trends.
Mathematical expression for that
Y = exp(b0) * exp (b1 * X )
Y = b2 * exp (b1 * X)
where b2 is the value of exp(b0)
In the polynomial, Tableau will make use of this expression for computing trends
Y = b0 + b1 *X + b2 * X ^ 2
In the power, Tableau will make use of this expression for computing trends
Y = b0 * X ^ b1
While Computing the trend lines in tableau on both ends , we need to take either date field or number field.
If we have only one textual dimensions on the view and if you want to add trend lines its not possible
In multi dimensional data sources , the trend lines are not allowed for dates because dates actually contain strings instead of numbers.
When we add trend line , tableau automatically creates a trend model.
It is always recommended to take the continuous dates while adding the trend line.
Trend lines are giving the output on the current and past data where as forecasting estimates future values.