How can I do regression analysis in Sage?
Question
I tried this unsuccessfully:
find_fit(data, quadratic_residues)
I am trying to find the best-fit for data about water flow rates: http://dl.getdropbox.com/u/175564/rate.png
---edit after the comment---
The new code:
var('x')
model(x) = x**2
find_fit((xlist, reqlist), model)
The error message:
Traceback (click to the left for traceback)
...
TypeError: data has to be a list of lists, a matrix, or a numpy array
---edit
The error message is now:
Traceback (click to the left for traceback)
...
ValueError: each row of data needs 2 entries, only 5 entries given
The same here as a picture: http://dl.getdropbox.com/u/175564/sage.png
Solution
mydata = [[1,3],[2,7],[3,13],[4,24]]
var('a,b,c')
mymodel(x) = a*x^2 + b*x + c
myfit = find_fit(mydata,mymodel,solution_dict=True)
points(mydata,color='purple') + plot(
mymodel(
a=myfit[a],
b=myfit[b],
c=myfit[c]
),
(x,0,4,),
color='red'
)
OTHER TIPS
I think your problem is that quadratic_residues probably doesn't mean what you think it means. If you are attempting to fit the best quadratic model I think you want to do something like.
var('a, b, c, x')
model(x) = a*x*x + b*x + c
find_fit(data, model)
Trying Steven his example I also ran into the error:
ValueError: each row of data needs 5 entries, only 2 entries given
Here is an more explicit example that I've tested to be working in sage 4.7.
sage: l=[4*i^2+7*i+134+random() for i in xrange(100)]
sage: var('a,b,c,x')
(a, b, c, x)
sage: model=a*x^2+b*x+c
sage: find_fit(zip(xrange(100),l),model,variables=[x])
[a == 4.0000723084513217, b == 6.9904742307159697, c == 134.74698715254667]
Apperently you need the variables=[x] to tell sage which of a,b,c and x corresponds to the variable in your model.