This app is a powerful data analysis tool. Data points can be added by typing, importing comma-quote delimited files, or through the calculator. It can produce PDF documents of the data analysis, graph and projected values and these can be printed or e-mailed. The PDF version of the graph can be converted to an image (bmp, jpeg, tiff, png) for inclusion in reports. Graphs can be generated in portrait or landscape orientation simply by rotating the device.
The scientific Reverse Polish Notation (RPN) calculator can be used for complex mathematical calculations. The calculator has hundreds of unit conversions built in to make complex calculations simple.
Use the Dataset Module to determine the “Best Fit” linear, polynomial, exponential or power equations for sets of data. It has the ability to store multiple datasets, calculate the goodness of fit, graph the given values verses the predicted values and calculate predicted values from the user’s input.
It is especially useful for scientists and science students because there are two modes of curve fitting: Calibration mode and Standard mode. In the Calibration Mode, an inverse prediction of X (abscissa) from Y (ordinate) is performed. As an example, if x is standard concentration values in mg/L and y is absorbance, then you can predict the concentration in mg/L from any absorbance value. Where polynomial fits are used, X is calculated using the equation roots. The standard mode is available to fit datasets where Y needs to be predicted from X or a high degree (up to 9th degree) polynomial fit is required.
Both modes provide the goodness of fit by calculating the correlation coefficient (r^2) and the relative percent difference (RPD) between the measured values and the predicted values.
For best results, when constructing a calibration curve space the standard values equally. This is to avoid high-leverage point error. For instance, if you perform a linear least squares fit of a set of data, the fitted line will always pass though the average X, average Y point. If the points are spaced equally then they will contribute equally to the fit. Also, do not use higher degree polynomials to compensate for detector saturation. In either mode, known values (for instance, standard concentration values) should be on the x-axis and the response variable (eg. instrument response measured with error) is on the y-axis. With this setup, the measurement errors are in the vertical direction and the least squares criterion works optimally.
Please note that the data is sorted by X values in the database and this affects the analysis and graphing functions. So, consider this when choosing modes.
Click for Instructions: CalcuCrvFitUserGuide