70.2K

Verified Solution

Link Copied!

# QUESTION $0---$ call in the data.

# CG Q$0$a # Read the data file bikeshare.csv into R

########## and name the object bikes. Use strings $=$ T$.$

# QUESTION $1---$ Data prep.

# CG Q$1$a # Run str$()$ on the bikes data frame to see

########## the quantitative variables and to confirm

########## the strings $=$ T argument read in character

########## data as a factor.

# CG Q$1$b # Run the following lines of code to make

########## month and weekdays, and holiday and work day

########## indicators factors.

bikes$mnth $<-$ factor$($bikes$mnth$)$

bikes$holiday $<-$ factor$($bikes$holiday$)$

bikes$weekday $<-$ factor$($bikes$weekday$)$

bikes$workingday $<-$ factor$($bikes$workingday$)$

# CG Q$1$c # Run str$()$ on the bikes data frame to see that

########## the Q$1$b code made these $4$ variables factors.

# QUESTION $2---$ Fit a regression model and inspect results

# CG Q$2$a # Fit a regression model called fit to model ride

########## count by temperature, humidity, windspeed,

########## weathersit, holiday indicator, and month.

# CG Q$2$b # Use the summary$()$ function to print the results of

########## the regression.

# CG Q$2$c # Use the coef$()$ function to print the

########## coefficient for September. Use the same

########## code you learned in previous regressions HWs$.$

# CG Q$2$d # Choose the correct interpretation.

########## A: As month increases by $1,$ expect count to decrease by $1265.$

########## B: As month increases by $1,$ expect count to increase by $1265.$

########## C: We expect count in September to be $1265$ lower than for January.

########## D: We expect count in September to be $1265$ higher than for January.

########## Use paste$()$ with the letter choice in quotes.

# CG Q$2$e # Calculate the R$-$squared using the formula from

########## previous regression HWs $($with deviance and null.deviance$).$

# CG Q$2$f # Find the autocorrelation at lag$-1$ for the residuals

########## from the regression. Use the code below.

acf$($fit$residuals, plot$=$F$)[1]$

# CG Q$2$g # Based on the results from Q$2$f$,$ is there autcorrelation

########## at lag$-1?$ Type paste$("$Y$")$ or paste$("$N$").$

# QUESTION $3---$ Fit an AR$1$ model to account for autcorrelation.

# CG Q$3$a # Run the provided code to add lagged count

########## and check that it worked as expected.

bikes$lag $<-$ c$($NA$,$ bikes$cnt$[2$:nrow$($bikes$)-1])$ #create lagged variable

head$($bikes$[,$c$("$cnt$",$"lag"$)])$ # confirm it worked as planned

# CG Q$3$b # Fit a regression using the variables from Q$2$a

########## with the addition of the lagged variable.

########## Call your fitted model ar$1$fit.

# CG Q$3$c # Find the R$-$squared using the same code

########## structure from previous HWs and lecture.

# CG Q$3$d # Use similar code to Q$2$f to print the

########## autocorrelation at lag $1$ for the residuals

########## from your AR regression.

# CG Q$3$e # Use the coef$()$ function to print the

########## coefficient for the lag$-1$ term.

# CG Q$3$f # Type paste$("$Y$")$ is the autocorrelation at

########## lag$-1$ is essentially $0.$ Otherwise, use paste$("$N$")$

# CG Q$3$g # The coefficient for the lag$-1$ term indicates

########## A: a random walk

########## B: stationary, mean$-$reverting

########## C: diverging

########## Use paste$()$ with the selected letter in quotes.

Answer & Explanation Solved by verified expert

See Answer