What is the extrapolation in graph?

Answer 1

Given x,y data are plotted on a graph and a trend to the data is identified (usually using regression), to extrapolate the data is to estimate y values beyond the known range, as shown on the x axis, of the data. In a more general sense, with multiple independent variables, an extrapolation would be going outside the known ranges of any of the independent variables in the prediction of the dependent variable.

Answer 2

You use extrapolation to estimate the value of a result (or point) outside the range of a series of your known values. It's like extending a "best fit line" on a graph, to see the hypothetical values if you had more data.

Example:

If you measure temperature and pressure in different situations and graph the points, it gives you a certain range of date, only the extremes in your date, but if you extrapolate, "extend the best fit line", you can see where it hits zero for example, which is helpful when dealing with "absolute zero."

Answer 3

Extrapolation is a prediction technique. Say that you collected data of the height of a baby when she was one year and 2 years old, as an example. The heights measured are 14 inches and 28 inches, respectively. How tall will the baby in her 3rd birthday? Simply connect the dots -- (1,14) and (2,28) -- if you plot the data on a graph paper, you can see what I mean. The units on the graph are years of age on the abscissa (x-axis) and inches in height in the ordinate (y-axis). This prediction technique is to extend that the line segment that connects these two points to 3 years on the horizontal (x) axis. The y-reading is, viola, 42 (inches). Extrapolation can be done also with non-linear curves, but in general, one just extrapolates linearly at the end of the data boundary; the slope is just taking what it is at the boundary. Microsoft Excel has the curve-fit function in scatterplot that one can use: 'add trendline.' The forecast option is available in the 'add trendline' section: 'forward' and 'backward' for a number of periods.

Looking at the graph, for certain, one can extrapolate the height to 20 years. The answer is 280 in or 23.3 ft! You can tell the fallacy or shortcoming of blindly extrapolating to beyond the validity of the technique with the given data of a short 2-year period. In general, the farther from the data range, the less certain is the prediction. There is no good substitute for good data.