> newdata <- read.csv(file="nr129717.csv", header=TRUE)
> head(newdata)
         y         x1       x2
1 19.15372 0.72292566 9.236719
2 11.13821 0.01474544 6.182988
3 18.06823 0.77878730 8.986456
4 12.08616 0.16159705 6.124310
5 16.72357 0.01543808 8.371611
6 10.68216 0.72488182 5.557159
> newy <- newdata[1]
> newx1 <- newdata[2]
> newx2 <- newdata[3]
> newy <- as.numeric(unlist(newy))
> newx1 <- as.numeric(unlist(newx1))
> newx2 <- as.numeric(unlist(newx2))
> l1 <- lm(newy~newx1+newx2)
> summary(l1)

Call:
lm(formula = newy ~ newx1 + newx2)

Residuals:
    Min      1Q  Median      3Q     Max
-6.1083 -0.9389 -0.0545  0.9359  6.3030

Coefficients:
            Estimate Std. Error t value Pr(>|t|)
(Intercept) -0.98126    0.27923  -3.514 0.000461 ***
newx1        0.94132    0.17442   5.397 8.48e-08 ***
newx2        2.01244    0.03499  57.520  < 2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 1.614 on 997 degrees of freedom
Multiple R-squared:  0.7706,    Adjusted R-squared:  0.7702
F-statistic:  1675 on 2 and 997 DF,  p-value: < 2.2e-16

> library(tseries)
> err <- residuals(l1)
> jarque.bera.test(err)

        Jarque Bera Test

data:  err
X-squared = 61.872, df = 2, p-value = 3.675e-14

> library(lmtest)
> dwtest(newy~newx1+newx2)

        Durbin-Watson test

data:  newy ~ newx1 + newx2
DW = 2.04, p-value = 0.7365
alternative hypothesis: true autocorrelation is greater than 0