Assignment 3
Alex De Leon
7/2/2021
Exercise 10
- Produce some numerical and graphical summaries of the Weekly data. Do there appear to be any patterns?
library(ISLR)
library(corrplot)## corrplot 0.90 loadedsummary(Weekly)## Year Lag1 Lag2 Lag3
## Min. :1990 Min. :-18.1950 Min. :-18.1950 Min. :-18.1950
## 1st Qu.:1995 1st Qu.: -1.1540 1st Qu.: -1.1540 1st Qu.: -1.1580
## Median :2000 Median : 0.2410 Median : 0.2410 Median : 0.2410
## Mean :2000 Mean : 0.1506 Mean : 0.1511 Mean : 0.1472
## 3rd Qu.:2005 3rd Qu.: 1.4050 3rd Qu.: 1.4090 3rd Qu.: 1.4090
## Max. :2010 Max. : 12.0260 Max. : 12.0260 Max. : 12.0260
## Lag4 Lag5 Volume Today
## Min. :-18.1950 Min. :-18.1950 Min. :0.08747 Min. :-18.1950
## 1st Qu.: -1.1580 1st Qu.: -1.1660 1st Qu.:0.33202 1st Qu.: -1.1540
## Median : 0.2380 Median : 0.2340 Median :1.00268 Median : 0.2410
## Mean : 0.1458 Mean : 0.1399 Mean :1.57462 Mean : 0.1499
## 3rd Qu.: 1.4090 3rd Qu.: 1.4050 3rd Qu.:2.05373 3rd Qu.: 1.4050
## Max. : 12.0260 Max. : 12.0260 Max. :9.32821 Max. : 12.0260
## Direction
## Down:484
## Up :605
##
##
##
## corrplot(cor(Weekly[,-9]), method="circle")
From looking at these summaries, it looks like there might be a linear relationship between year and volume.
- Use the full data set to perform a logistic regression with Direction as the response and the five lag variables plus Volume as predictors. Use the summary() function to print the results. Do any of the predictors appear to be statistically significant? If so,which ones?
attach(Weekly)
part10b_model <- glm(Direction ~ Lag1 + Lag2 + Lag3 + Lag4 + Lag5 + Volume, data = Weekly, family = binomial)
summary(part10b_model)##
## Call:
## glm(formula = Direction ~ Lag1 + Lag2 + Lag3 + Lag4 + Lag5 +
## Volume, family = binomial, data = Weekly)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.6949 -1.2565 0.9913 1.0849 1.4579
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 0.26686 0.08593 3.106 0.0019 **
## Lag1 -0.04127 0.02641 -1.563 0.1181
## Lag2 0.05844 0.02686 2.175 0.0296 *
## Lag3 -0.01606 0.02666 -0.602 0.5469
## Lag4 -0.02779 0.02646 -1.050 0.2937
## Lag5 -0.01447 0.02638 -0.549 0.5833
## Volume -0.02274 0.03690 -0.616 0.5377
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1496.2 on 1088 degrees of freedom
## Residual deviance: 1486.4 on 1082 degrees of freedom
## AIC: 1500.4
##
## Number of Fisher Scoring iterations: 4The only variable that seems to be statistically significant is Lag2 with the p-value of 0.0296.
- Compute the confusion matrix and overall fraction of correct predictions. Explain what the confusion matrix is telling you about the types of mistakes made by logistic regression.
partc.prob= predict(part10b_model, type='response')
partc.pred =rep("Down", length(partc.prob))
partc.pred[partc.prob > 0.5] = "Up"
table(partc.pred, Direction)## Direction
## partc.pred Down Up
## Down 54 48
## Up 430 557## Calculation to find percentage of current current conditions.
(54+557)/(54+48+430+557)## [1] 0.5610652This model predicted 56.11% of the trend correctly.
- Now fit the logistic regression model using a training data period from 1990 to 2008, with Lag2 as the only predictor. Compute the confusion matrix and the overall fraction of correct predictions for the held out data (that is, the data from 2009 and 2010).
train = (Year<2009)
weekly_partd_model <-Weekly[!train,]
partd_model<-glm(Direction~Lag2, data=Weekly,family=binomial, subset=train)
partd.prob= predict(partd_model, weekly_partd_model, type = "response")
partd.pred = rep("Down", length(partd.prob))
partd.pred[partd.prob > 0.5] = "Up"
Direction.data = Direction[!train]
table(partd.pred, Direction.data)## Direction.data
## partd.pred Down Up
## Down 9 5
## Up 34 56mean(partd.pred == Direction.data)## [1] 0.625This yields an accuracy of 62.5%.
- Repeat (d) using LDA.
library(MASS)
lda_model<-lda(Direction~Lag2, data=Weekly,family=binomial, subset=train)
lda.pred<-predict(lda_model, weekly_partd_model)
table(lda.pred$class, Direction.data)## Direction.data
## Down Up
## Down 9 5
## Up 34 56mean(lda.pred$class==Direction.data)## [1] 0.625This yields 62.5% accuracy.
- Repeat (d) using QDA.
qda_model = qda(Direction ~ Lag2, data = Weekly, subset = train)
qda.pred = predict(qda_model, weekly_partd_model)$class
table(qda.pred, Direction.data)## Direction.data
## qda.pred Down Up
## Down 0 0
## Up 43 61mean(qda.pred==Direction.data)## [1] 0.5865385This yields 58.65% accuracy.
- Repeat (d) using KNN with K = 1.
library(class)
week_train = as.matrix(Lag2[train])
week_test = as.matrix(Lag2[!train])
train_direction = Direction[train]
set.seed(1)
knn.pred = knn(week_train, week_test, train_direction, k=1)
table(knn.pred, Direction.data)## Direction.data
## knn.pred Down Up
## Down 21 30
## Up 22 31mean(knn.pred == Direction.data)## [1] 0.5This yields an accuracy of 50%.
- Which of these methods appears to provide the best results on this data?
Of these methods Logistic Regression and LDA yield the highest accuracies.
- Experiment with different combinations of predictors, including possible transformations and interactions, for each of the methods. Report the variables, method, and associated confusion matrix that appears to provide the best results on the held out data. Note that you should also experiment with values for K in the KNN classifier.
## Regression with Interaction Lag1 and Lag2
part_j_model<-glm(Direction ~ Lag1:Lag2 + Lag2, data=Weekly,family=binomial, subset=train)
partj.prob= predict(part_j_model, weekly_partd_model, type = "response")
partj.pred = rep("Down", length(partj.prob))
partj.pred[partj.prob > 0.5] = "Up"
Direction.data = Direction[!train]
table(partj.pred, Direction.data)## Direction.data
## partj.pred Down Up
## Down 3 3
## Up 40 58## LDA Interaction with Lag1:Lag2
ldaint_model<-lda(Direction~Lag1:Lag2+Lag1, data=Weekly,family=binomial, subset=train)
ldaint.pred<-predict(ldaint_model, weekly_partd_model)
table(ldaint.pred$class, Direction.data)## Direction.data
## Down Up
## Down 4 4
## Up 39 57mean(ldaint.pred$class==Direction.data)## [1] 0.5865385## QDA Interaction with Lag1:Lag2
qdaint_model<-qda(Direction~Lag1:Lag2+Lag1, data=Weekly,family=binomial, subset=train)
qdaint.pred<-predict(qdaint_model, weekly_partd_model)
table(qdaint.pred$class, Direction.data)## Direction.data
## Down Up
## Down 19 36
## Up 24 25## k = 35
knn_train=as.matrix(Lag2[train])
knn_test=as.matrix(Lag2[!train])
knn_train_direction =Direction[train]
set.seed(1)
knn35.pred=knn(knn_train, knn_test,knn_train_direction,k=10)
table(knn35.pred,Direction.data)## Direction.data
## knn35.pred Down Up
## Down 17 21
## Up 26 40mean(knn35.pred == Direction.data)## [1] 0.5480769Exercise 11
- Create a binary variable, mpg01, that contains a 1 if mpg contains a value above its median, and a 0 if mpg contains a value below its median. You can compute the median using the median() function. Note you may find it helpful to use the data.frame() function to create a single data set containing both mpg01 and the other Auto variables.
library(ISLR)
attach(Auto)
summary(Auto)## mpg cylinders displacement horsepower weight
## Min. : 9.00 Min. :3.000 Min. : 68.0 Min. : 46.0 Min. :1613
## 1st Qu.:17.00 1st Qu.:4.000 1st Qu.:105.0 1st Qu.: 75.0 1st Qu.:2225
## Median :22.75 Median :4.000 Median :151.0 Median : 93.5 Median :2804
## Mean :23.45 Mean :5.472 Mean :194.4 Mean :104.5 Mean :2978
## 3rd Qu.:29.00 3rd Qu.:8.000 3rd Qu.:275.8 3rd Qu.:126.0 3rd Qu.:3615
## Max. :46.60 Max. :8.000 Max. :455.0 Max. :230.0 Max. :5140
##
## acceleration year origin name
## Min. : 8.00 Min. :70.00 Min. :1.000 amc matador : 5
## 1st Qu.:13.78 1st Qu.:73.00 1st Qu.:1.000 ford pinto : 5
## Median :15.50 Median :76.00 Median :1.000 toyota corolla : 5
## Mean :15.54 Mean :75.98 Mean :1.577 amc gremlin : 4
## 3rd Qu.:17.02 3rd Qu.:79.00 3rd Qu.:2.000 amc hornet : 4
## Max. :24.80 Max. :82.00 Max. :3.000 chevrolet chevette: 4
## (Other) :365mpg01 <- rep(0, length(mpg))
mpg01[mpg > median(mpg)] <- 1
Auto = data.frame(Auto, mpg01)- Explore the data graphically in order to investigate the association between mpg01 and the other features. Which of the other features seem most likely to be useful in predicting mpg01? Scatterplots and boxplots may be useful tools to answer this question.Describe your findings.
corrplot(cor(Auto[,-9]), method="circle")
Cylinders, Displacement, and Weight look like they all strongly correlate negatively with mpg01. Horsepower also seemes to negatibely correlate.
- Split the data into a training set and a test set.
train_split <- (year %% 2 == 0)
train_auto <- Auto[train_split,]
test_auto <- Auto[-train_split,]- Perform LDA on the training data in order to predict mpg01 using the variables that seemed most associated with mpg01 in (b). What is the test error of the model obtained?
autolda <- lda(mpg01~displacement+horsepower+weight+year+cylinders+origin, data=train_auto)
autolda_pred <- predict(autolda, test_auto)
table(autolda_pred$class, test_auto$mpg01)##
## 0 1
## 0 169 7
## 1 26 189mean(autolda_pred$class != test_auto$mpg01)## [1] 0.08439898The test error is 8.44%.
- Perform QDA on the training data in order to predict mpg01 using the variables that seemed most associated with mpg01 in (b). What is the test error of the model obtained?
autoqda <- qda(mpg01~displacement+horsepower+weight+year+cylinders+origin, data=train_auto)
autoqda_pred <- predict(autoqda, test_auto)
table(autoqda_pred$class, test_auto$mpg01)##
## 0 1
## 0 176 20
## 1 19 176mean(autoqda_pred$class != test_auto$mpg01)## [1] 0.09974425The test error is 9.97%.
- Perform logistic regression on the training data in order to predict mpg01 using the variables that seemed most associated with mpg01 in (b). What is the test error of the model obtained?
auto_log_model<-glm(mpg01~displacement+horsepower+weight+year+cylinders+origin, data=train_auto,family=binomial)
auto_probs = predict(auto_log_model, test_auto, type = "response")
auto_pred = rep(0, length(auto_probs))
auto_pred[auto_probs > 0.5] = 1
table(auto_pred, test_auto$mpg01)##
## auto_pred 0 1
## 0 174 12
## 1 21 184mean(auto_pred != test_auto$mpg01)## [1] 0.08439898The test error rate is 8.44%.
- Perform KNN on the training data, with several values of K, in order to predict mpg01. Use only the variables that seemed most associated with mpg01 in (b). What test errors do you obtain? Which value of K seems to perform the best on this data set?
#K=5
#train_k5= cbind(displacement,horsepower,weight,cylinders,year,origin)[train,]
#test_k5=cbind(displacement,horsepower,weight,cylinders, year, origin)[-train,]
#set.seed(1)
#autok5_pred=knn(train_k5,test_k5,train_auto$mpg01,k=5)
#mean(autok5_pred != test_auto$mpg01)
## I tried running this chunkk but kept getting errors. Commented it out so i could knit documentExercise 13
Using the Boston data set, fit classification models in order to predict whether a given suburb has a crime rate above or below the median.Explore logistic regression, LDA, and KNN models using various subsets of the predictors. Describe your findings.
summary(Boston)## crim zn indus chas
## Min. : 0.00632 Min. : 0.00 Min. : 0.46 Min. :0.00000
## 1st Qu.: 0.08205 1st Qu.: 0.00 1st Qu.: 5.19 1st Qu.:0.00000
## Median : 0.25651 Median : 0.00 Median : 9.69 Median :0.00000
## Mean : 3.61352 Mean : 11.36 Mean :11.14 Mean :0.06917
## 3rd Qu.: 3.67708 3rd Qu.: 12.50 3rd Qu.:18.10 3rd Qu.:0.00000
## Max. :88.97620 Max. :100.00 Max. :27.74 Max. :1.00000
## nox rm age dis
## Min. :0.3850 Min. :3.561 Min. : 2.90 Min. : 1.130
## 1st Qu.:0.4490 1st Qu.:5.886 1st Qu.: 45.02 1st Qu.: 2.100
## Median :0.5380 Median :6.208 Median : 77.50 Median : 3.207
## Mean :0.5547 Mean :6.285 Mean : 68.57 Mean : 3.795
## 3rd Qu.:0.6240 3rd Qu.:6.623 3rd Qu.: 94.08 3rd Qu.: 5.188
## Max. :0.8710 Max. :8.780 Max. :100.00 Max. :12.127
## rad tax ptratio black
## Min. : 1.000 Min. :187.0 Min. :12.60 Min. : 0.32
## 1st Qu.: 4.000 1st Qu.:279.0 1st Qu.:17.40 1st Qu.:375.38
## Median : 5.000 Median :330.0 Median :19.05 Median :391.44
## Mean : 9.549 Mean :408.2 Mean :18.46 Mean :356.67
## 3rd Qu.:24.000 3rd Qu.:666.0 3rd Qu.:20.20 3rd Qu.:396.23
## Max. :24.000 Max. :711.0 Max. :22.00 Max. :396.90
## lstat medv
## Min. : 1.73 Min. : 5.00
## 1st Qu.: 6.95 1st Qu.:17.02
## Median :11.36 Median :21.20
## Mean :12.65 Mean :22.53
## 3rd Qu.:16.95 3rd Qu.:25.00
## Max. :37.97 Max. :50.00attach(Boston)Creating a binary variable from crime:
crime01 <- rep(0, length(crim))
crime01[crim > median(crim)] <- 1
Boston= data.frame(Boston,crime01)Splitting the data into Train and Test sets:
train = 1:(dim(Boston)[1]/2)
test = (dim(Boston)[1]/2 + 1):dim(Boston)[1]
train_boston = Boston[train, ]
test_boston = Boston[test, ]
crime01_test = crime01[test]Graph of Correlating variables:
corrplot(cor(Boston), method="circle")
Logistic Regression
set.seed(100)
boston_log <-glm(crime01~ indus+nox+age+dis+rad+tax, data= train_boston,family=binomial)
boston_probs = predict(boston_log, test_boston, type = "response")
boston_pred = rep(0, length(boston_probs))
boston_pred[boston_probs > 0.5] = 1
table(boston_pred, crime01_test)## crime01_test
## boston_pred 0 1
## 0 75 8
## 1 15 155mean(boston_pred != crime01_test)## [1] 0.09090909LDA
boston_lda <-lda(crime01~ indus+nox+age+dis+rad+tax, data=train_boston,family=binomial)
bostonlda_pred = predict(boston_lda, test_boston)
table(bostonlda_pred$class, crime01_test)## crime01_test
## 0 1
## 0 81 18
## 1 9 145mean(bostonlda_pred$class != crime01_test)## [1] 0.1067194K Nearest Neighbors
### k = 10
train_k10=cbind(indus,nox,age,dis,rad,tax)[train,]
test_k10=cbind(indus,nox,age,dis,rad,tax)[test,]
boston_knn_pred=knn(train_k10, test_k10, crime01_test, k=10)
table(boston_knn_pred,crime01_test)## crime01_test
## boston_knn_pred 0 1
## 0 39 10
## 1 51 153mean(boston_knn_pred !=crime01_test)## [1] 0.2411067### k = 20
train_k20=cbind(indus,nox,age,dis,rad,tax)[train,]
test_k20=cbind(indus,nox,age,dis,rad,tax)[test,]
boston_knn_pred20=knn(train_k20, test_k20, crime01_test, k=20)
table(boston_knn_pred20,crime01_test)## crime01_test
## boston_knn_pred20 0 1
## 0 41 17
## 1 49 146mean(boston_knn_pred20 !=crime01_test)## [1] 0.2608696### k = 50
train_k50=cbind(indus,nox,age,dis,rad,tax)[train,]
test_k50=cbind(indus,nox,age,dis,rad,tax)[test,]
boston_knn_pred50=knn(train_k20, test_k50, crime01_test, k=50)
table(boston_knn_pred50,crime01_test)## crime01_test
## boston_knn_pred50 0 1
## 0 38 13
## 1 52 150mean(boston_knn_pred50 !=crime01_test)## [1] 0.256917By examining all of the models, the Logistic regression yielded the lowest test error of 9.09%. The KNN models yielded the highest error, with the error rising as we increase k.