Open eamonnmag opened 8 years ago
This issue was a suggestion from Gavin Salam:
- Suppose you have an entry with 10 data points. First of all, each data point should get an label, e.g. 0..9.
- Then suppose you have a model that’s expected to describe the data over the phase space defined by points 4..8.
- the information that people usually want to get is the compatibility of their model with those data points
- so ideally one would like to have a program that allows the user to pass their “prediction” for those data points and get back a chi^2.
Some more detail for a suggested implementation: make a button "Compare" below each dependent variable column if data points are numerical and have errors. Clicking button gives pop-up box prompting user to upload predictions, say, in a CSV format with two columns (index of data point and numerical value of prediction). Code calculates the chi^2 defined as the sum over data points of [(data - theory)/error]^2, where the denominator is simply the sum in quadrature of all individual errors. The chi^2 is printed out and ideally the predictions are also added to the plot.
Calculate Chi squared on an uploaded input vs some given data.