09-05-2018, 10:48 PM
The uncertainties should be no problem. I am not sure if many users are able to interpret them correctly (there are so many ways to misinterpret these even for a linear fit), but I will add them.
(This reminds me of students who did a linear fit on temperatures in the range of 10°C to 80°C to find at which point they cross 0°C. The results were quite good, but they got an error of +/-1000°C because they did their calculations in Kelvin. Understanding why the error was much smaller when doing the same math in °C took a while...)
Selecting the range is a bit more difficult. If we just use the zoom, should the fit be updated when the user tries to zoom into the fit results? Currently the fit also updates as new data arrives. This is very useful for the centripetal acceleration experiment, but how should this be handled when the zoom follows new data. I think the range should be independent of the zoom, but I am afraid that this is a bigger task which we should should tackle in a later update. The tool-menu will certainly grow in the future...
(This reminds me of students who did a linear fit on temperatures in the range of 10°C to 80°C to find at which point they cross 0°C. The results were quite good, but they got an error of +/-1000°C because they did their calculations in Kelvin. Understanding why the error was much smaller when doing the same math in °C took a while...)
Selecting the range is a bit more difficult. If we just use the zoom, should the fit be updated when the user tries to zoom into the fit results? Currently the fit also updates as new data arrives. This is very useful for the centripetal acceleration experiment, but how should this be handled when the zoom follows new data. I think the range should be independent of the zoom, but I am afraid that this is a bigger task which we should should tackle in a later update. The tool-menu will certainly grow in the future...