The forestfires.csv data base contains meteorological information and the area burned for $517$ forest fires that occurred in Montesinho Natural Park in Portugal. The columns of interest are FFMC$,$ DMC$,$ DC$,$ and ISI, as well as the response variable, area.

Read in the file forestfires.csv$.$

Use the ols$()$ function to perform forward selection stepwise regression on the variables FFMC$,$ DMC$,$ DC$,$ and ISI, in that order.

Continue the stepwise regression until no new variables have a p$-$value less than $0\mathrm{.}15.$

Ex: If the variable RH is used instead of FFMC$,$ the output is:

model$1\_1$ p$-$values are Intercept $1\mathrm{.}384309$e$-96$

RH $1\mathrm{.}577449$e$-01$

dtype: float$64$

model$1\_2$ p$-$values are Intercept $3\mathrm{.}710545$e$-158$

DMC $8\mathrm{.}598943$e$-01$

dtype: float$64$

model$1\_3$ p$-$values are Intercept $3\mathrm{.}005972$e$-138$

DC $2\mathrm{.}139832$e$-02$

dtype: float$64$

model$1\_4$ p$-$values are Intercept $4\mathrm{.}580488$e$-144$

ISI $5\mathrm{.}785291$e$-01$

dtype: float$64$

model$2\_1$ p$-$values are Intercept $1\mathrm{.}284725$e$-75$

RH $1\mathrm{.}838245$e$-01$

DC $2\mathrm{.}450738$e$-02$

dtype: float$64$

model$2\_2$ p$-$values are Intercept $2\mathrm{.}969390$e$-136$

DMC $1\mathrm{.}642624$e$-02$

DC $9\mathrm{.}065366$e$-04$

dtype: float$64$

model$2\_3$ p$-$values are Intercept $1\mathrm{.}656648$e$-113$

DC $2\mathrm{.}569129$e$-02$

ISI $9\mathrm{.}756819$e$-01$

dtype: float$64$

model$3\_1$ p$-$values are Intercept $3\mathrm{.}247158$e$-76$

RH $3\mathrm{.}108264$e$-01$

DMC $2\mathrm{.}520680$e$-02$

DC $1\mathrm{.}521100$e$-03$

dtype: float$64$

model$3\_2$ p$-$values are Intercept $3\mathrm{.}674604$e$-114$

DMC $1\mathrm{.}397040$e$-02$

DC $9\mathrm{.}735041$e$-04$

ISI $5\mathrm{.}855461$e$-01$

dtype: float$64$

# import the necessary modules

fires $=$ # read in the csv file

# response variable

Y $=$ # set area as the response variable

# first forward selection step

model$1\_1=$ # generate the linear regression model for FFMC

# prints the p$-$value

print$("$model$1\_1$ p$-$values are $",$ model$1\_1.$pvalues$)$

model$1\_2=$ # generate the linear regression model for wind

print$("$model$1\_2$ p$-$values are $",$ model$1\_2.$pvalues$)$

model$1\_3=$ # generate the linear regression model for DC

print$("$model$1\_3$ p$-$values are $",$ model$1\_3.$pvalues$)$

model$1\_4=$ # generate the linear regression model for temp

print$("$model$1\_4$ p$-$values are $",$ model$1\_4.$pvalues$)$

# Second forward selection step using order FFMC$,$ wind, DC$,$ temp. Ex: If the best SLR model uses temp, the first MLR model should be Y ~ FFMC $+$ temp.

model$2\_1=$ # generate multiple regression model

print$("$model$2\_1$ p$-$values are $",$ model$2\_1.$pvalues$)$

model$2\_2=$ # generate multiple regression model

print$("$model$2\_2$ p$-$values are $",$ model$2\_2.$pvalues$)$

model$2\_3=$ # generate multiple regression model

print$("$model$2\_3$ p$-$values are $",$ model$2\_3.$pvalues$)$

# Third forward regression step using order FFMC$,$ wind, DC$,$ temp

model$3\_1=$ # generate multiple regression model

print$("$model$3\_1$ p$-$values are $",$ model$3\_1.$pvalues$)$

model$3\_2=$ # generate multiple regression model

print$("$model$3\_2$ p$-$values are $",$ model$3\_2.$pvalues$)$

# Continue until no additional predictor variables have p$-$values less than $0\mathrm{.}15.$