CODEWITHSUNDEEP
pythonpython-3.xregressiondata-fitting
Amarth Gûl
Amarth Gûl

Reputation: 1080

How to build a regression model in python?

I want to build a regression model for a data set, already know that :

x_1 and y is quadratic relationship, x_2 and y is linear; but not sure whether x_2 has quadratic relationship with y, nor if x_1 and x_2 has some sort of interaction. 

x_1: ['66.29', '40.96', '73.00', '45.01', '57.20', '26.85', '38.12', '35.84', '75.80', '37.41', '54.38', '46.19', '46.13', '30.37', '39.06', '79.38', '52.77', '55.92']
x_2: ['7.00', '5.00', '10.00', '6.00', '4.00', '5.00', '4.00', '6.00', '9.00', '5.00', '2.00', '7.00', '4.00', '3.00', '5.00', '1.00', '8.00', '6.00'] 
y: ['196.00', '63.00', '252.00', '84.00', '126.00', '14.00', '49.00', '49.00', '266.00', '49.00', '105.00', '98.00', '77.00', '14.00', '56.00', '245.00', '133.00', '133.00']

So I constructed that function:SampleFunction

But I don't know how to evaluate it, I tried curve_fit in scipy, yet seems it does not work for multiple independent variables. So is there a way to do that in python?

Upvotes: 1

Views: 649

Answers (1)

Pankaj
Pankaj

Reputation: 517

Sckit-learn package in python includes both linear and polynomial regression models. Have a look at the link : linear and polynomial regression models.

Basically, y = c1 + c2 * x1 + c3 * x2 + c4 * x1^2 + c5 * x2^2 + c6 * x1 * x2 can be transformed by defining new variable z = [x1, x2, x1^2, x2^2, x1*x2].

With this transformation, the equation can be rewritten as

y = c1 + c2 z1 + c3 * z2 + c4 * z3 + c5 * z4 + c6 * z5.

Thus, the problem of polynomial fitting has now been reduced to linear one and the linear model trained on polynomial features is able to exactly recover the input polynomial coefficients.

You can find several examples of polynomial regression in the link above.

Upvotes: 1

Related Questions