How do I improve my multiple regression model?

I am attaching my code, if you look at my coefficents they are huge and even my max error. How can I improve my results?

T = readtable('a.csv');
x1 = T.T;
x2 = T.Hour;
x3 = T.is_holiday;
x4 = T.month;
y = T.load;

A model of this data looks like :

Multiple regressions solves for coefficents  by using a least squares fit. 
Following line forms a matrix , X which has the first row set as 1, as it is necessary to do so in stats to have a constant term. 
X = [ones(size(x1)) x1 x2 x3];
Solve for the different coefficents. 

a = X\y

2.7423e+03\\ 0.8635 \\ 42.1795 \\ -87.1325 \\

The least- square fit model of the data is y = 27380+0.9x_1+42.2x_2-87.1x_3

Y = X*a;
MaxErr = max(abs(Y-y))

[KSSV]

Did you try polyfit and curve fitting toolbox?