Final Project
Owura Kuffuor
December 4, 2017
library(tidyverse)
library(car)
library(janitor)
library(ggcorrplot)
library(dotwhisker)
bball <- read_csv("https://raw.githubusercontent.com/ryanburge/pls2003_sp17/master/bball.csv")What team won the most games in a season in the dataset? What year was that?
bball %>% select(team_id,w,year) %>% arrange(-w)## # A tibble: 854 x 3
## team_id w year
## <chr> <int> <int>
## 1 SEA 116 2001
## 2 NYA 114 1998
## 3 NYN 108 1986
## 4 ATL 106 1998
## 5 SLN 105 2004
## 6 ATL 104 1993
## 7 OAK 104 1988
## 8 ATL 103 1999
## 9 NYA 103 2002
## 10 NYA 103 2009
## # ... with 844 more rowsWhat did scored the most runs in a single season? What year was that?
bball %>% select(team_id,r,year) %>% arrange(-r)## # A tibble: 854 x 3
## team_id r year
## <chr> <int> <int>
## 1 CLE 1009 1999
## 2 SEA 993 1996
## 3 CHA 978 2000
## 4 COL 968 2000
## 5 NYA 968 2007
## 6 NYA 965 1998
## 7 BOS 961 2003
## 8 COL 961 1996
## 9 CLE 952 1996
## 10 CLE 950 2000
## # ... with 844 more rowsWhat team hit the most home runs + triples (combined) in a single season. What year was that?
bball<-bball %>% mutate(hrtriple=hr+triple)
bball %>% select(team_id,hrtriple,year) %>% arrange(-hrtriple)## # A tibble: 854 x 3
## team_id hrtriple year
## <chr> <int> <int>
## 1 TEX 289 2005
## 2 BAL 286 1996
## 3 HOU 285 2000
## 4 SEA 285 1997
## 5 COL 279 1997
## 6 BOS 278 2003
## 7 TOR 278 2010
## 8 SFN 275 2001
## 9 TEX 275 2003
## 10 COL 274 2001
## # ... with 844 more rowsQuestion 2
What team averaged the most wins per season in the dataset?
bball %>% group_by(team_id) %>% summarise(meanw=mean(w))%>% arrange(-meanw)## # A tibble: 34 x 2
## team_id meanw
## <chr> <dbl>
## 1 LAA 90.09091
## 2 NYA 89.90000
## 3 ATL 86.36667
## 4 BOS 85.76667
## 5 SLN 85.26667
## 6 OAK 84.13333
## 7 SFN 83.76667
## 8 LAN 83.40000
## 9 TOR 81.76667
## 10 NYN 81.40000
## # ... with 24 more rowsWhich league (AL or NL) has a higher level of offense since 2000?
bball %>% filter(league_id=="AL") %>% filter(year>1999) %>% group_by(league_id) %>% summarise(meanr=mean(r))## # A tibble: 1 x 2
## league_id meanr
## <chr> <dbl>
## 1 AL 760.8108bball %>% filter(league_id=="NL") %>% filter(year>1999) %>% group_by(league_id) %>% summarise(meanr=mean(r))## # A tibble: 1 x 2
## league_id meanr
## <chr> <dbl>
## 1 NL 719.7628Does one league pay their player better than the other league since 2005?
bball %>% filter(league_id=="NL") %>% filter(year>=2005) %>% group_by(league_id) %>% summarise(meanr=mean(salary)) ## # A tibble: 1 x 2
## league_id meanr
## <chr> <dbl>
## 1 NL 3218701bball %>% filter(league_id=="AL") %>% filter(year>=2005) %>% group_by(league_id) %>% summarise(meanr=mean(salary))## # A tibble: 1 x 2
## league_id meanr
## <chr> <dbl>
## 1 AL 3578627Question 3
Is there a relationship between amount a team pays in salary and how many wins they achieve? Visualize this result with points and a line.
bball %>% summarise(cor(w, salary))## # A tibble: 1 x 1
## `cor(w, salary)`
## <dbl>
## 1 0.269302 bball %>% ggplot(.,aes(x=w, y=salary)) + geom_point()+ geom_smooth(method = lm)
Is there a relationship between amount of wins a team achieves and their attendance for the year? Again, visualize the result.
bball %>% summarise(cor(w, attendance))## # A tibble: 1 x 1
## `cor(w, attendance)`
## <dbl>
## 1 0.4959296bball %>% ggplot(.,aes(x=w, y=attendance)) + geom_point()+ geom_smooth(method = lm)
Does this relationship (between wins and attendance) change if the team is in the American league vs the National League? Visualize this.
bball %>% filter(league_id=="NL") %>% summarise(cor(w, attendance))## # A tibble: 1 x 1
## `cor(w, attendance)`
## <dbl>
## 1 0.4536521bball %>% filter(league_id=="NL") %>% ggplot(.,aes(x=w, y=attendance)) + geom_point()+ geom_smooth(method = lm)
bball %>% filter(league_id=="AL") %>% summarise(cor(w, attendance))## # A tibble: 1 x 1
## `cor(w, attendance)`
## <dbl>
## 1 0.5497723bball %>% filter(league_id=="AL") %>% ggplot(.,aes(x=w, y=attendance)) + geom_point()+ geom_smooth(method = lm)
bball %>% group_by(league_id) %>% summarise(cor(w, attendance))## # A tibble: 2 x 2
## league_id `cor(w, attendance)`
## <chr> <dbl>
## 1 AL 0.5497723
## 2 NL 0.4536521bball %>% ggplot(.,aes(x=w, y=attendance, color= as.factor(league_id))) + geom_point() + geom_smooth(method = lm)
Question 4
house <- read_csv("https://raw.githubusercontent.com/ryanburge/pls2003_sp17/master/house.csv")Construct a regression model that uses “price” as the dependent variable. One independent variable needs to be the age of home. The other four can be your choice.
house<-house %>% mutate(ageofhome=2017-yr_built)
reg1 <- lm(price ~ ageofhome + bedrooms + bathrooms+ condition + floors, data = house)
summary(reg1)##
## Call:
## lm(formula = price ~ ageofhome + bedrooms + bathrooms + condition +
## floors, data = house)
##
## Residuals:
## Min 1Q Median 3Q Max
## -1660737 -167690 -33282 108864 5656240
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -440278.7 15596.7 -28.229 < 2e-16 ***
## ageofhome 3761.8 88.7 42.409 < 2e-16 ***
## bedrooms 5598.1 2580.6 2.169 0.0301 *
## bathrooms 298416.0 3769.5 79.166 < 2e-16 ***
## condition 17444.2 3386.6 5.151 2.61e-07 ***
## floors 65591.0 4623.5 14.186 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 297900 on 21607 degrees of freedom
## Multiple R-squared: 0.3425, Adjusted R-squared: 0.3424
## F-statistic: 2252 on 5 and 21607 DF, p-value: < 2.2e-16Intepret the regression output.
The results show that all the variables used are statiscally significant.With every addition of age of th ehouse, the value of the price increases by 3761.8. An addition of a bedroom leads to th eincrease of price by 55981 dollars. Also the addition of a bathroom leas to an increase in the value of th ehouse by 298416.0. An increase in the condition of the house by one point leads to increment in price by 17444.2 dollars. Finally, an increase in the number of floors leads to an uncrease in price of the house by 65591.0 dollars.
Question 5
Making a dotwhisker plot of the regression analysis you just conducted.
dwplot(reg1) + geom_vline(xintercept = 0, colour = "grey50", linetype = 2)
Now, scale your variables properly and make that dotwhisker plot.
house<-house %>% mutate(bd = bedrooms/33, bth = bathrooms/8, age = ageofhome/117, cd=condition/5, fl=floors/3.5)
reg1 <- lm(price ~ age + bd + bth + cd + fl, data = house)
summary(reg1)##
## Call:
## lm(formula = price ~ age + bd + bth + cd + fl, data = house)
##
## Residuals:
## Min 1Q Median 3Q Max
## -1660737 -167690 -33282 108864 5656240
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -440279 15597 -28.229 < 2e-16 ***
## age 440135 10378 42.409 < 2e-16 ***
## bd 184739 85160 2.169 0.0301 *
## bth 2387328 30156 79.166 < 2e-16 ***
## cd 87221 16933 5.151 2.61e-07 ***
## fl 229569 16182 14.186 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 297900 on 21607 degrees of freedom
## Multiple R-squared: 0.3425, Adjusted R-squared: 0.3424
## F-statistic: 2252 on 5 and 21607 DF, p-value: < 2.2e-16dwplot(reg1) + geom_vline(xintercept = 0, colour = "grey50", linetype = 2)