BUA451 Conditional Tree

data <- read.csv("MVcat.csv")

library(caret)

## 载入需要的程序包：ggplot2

## 载入需要的程序包：lattice

library(party)

## 载入需要的程序包：grid

## 载入需要的程序包：mvtnorm

## 载入需要的程序包：modeltools

## 载入需要的程序包：stats4

## 载入需要的程序包：strucchange

## 载入需要的程序包：zoo

## 
## 载入程序包：'zoo'

## The following objects are masked from 'package:base':
## 
##     as.Date, as.Date.numeric

## 载入需要的程序包：sandwich

data$MVcat <- as.factor(data$MVcat)
data$HouseAge <- as.numeric(data$HouseAge)
data$SquareFeet <- as.numeric(data$SquareFeet)

set.seed(1234)

sample_size <- floor(0.7 * nrow(data))
train_indices <- sample(seq_len(nrow(data)), size = sample_size, replace = FALSE)
training_data <- data[train_indices, ]
testing_data <- data[-train_indices, ]

ctree_model <- ctree(MVcat ~ HouseAge + SquareFeet, data = training_data)
predictions <- predict(ctree_model, newdata = testing_data)

confusion_matrix <- confusionMatrix(predictions, testing_data$MVcat)
summary(confusion_matrix)

##          Length Class  Mode     
## positive  0     -none- NULL     
## table    25     table  numeric  
## overall   7     -none- numeric  
## byClass  55     -none- numeric  
## mode      1     -none- character
## dots      0     -none- list

print(confusion_matrix)

## Confusion Matrix and Statistics
## 
##              Reference
## Prediction    AVE HIGH LOW MEDIUM HIGH MEDIUM LOW
##   AVE           0    0   0           0          0
##   HIGH          0    0   0           0          0
##   LOW           1    0   3           0          0
##   MEDIUM HIGH   2    3   0           1          3
##   MEDIUM LOW    0    0   0           0          0
## 
## Overall Statistics
##                                           
##                Accuracy : 0.3077          
##                  95% CI : (0.0909, 0.6143)
##     No Information Rate : 0.2308          
##     P-Value [Acc > NIR] : 0.3515          
##                                           
##                   Kappa : 0.2095          
##                                           
##  Mcnemar's Test P-Value : NA              
## 
## Statistics by Class:
## 
##                      Class: AVE Class: HIGH Class: LOW Class: MEDIUM HIGH
## Sensitivity              0.0000      0.0000     1.0000            1.00000
## Specificity              1.0000      1.0000     0.9000            0.33333
## Pos Pred Value              NaN         NaN     0.7500            0.11111
## Neg Pred Value           0.7692      0.7692     1.0000            1.00000
## Prevalence               0.2308      0.2308     0.2308            0.07692
## Detection Rate           0.0000      0.0000     0.2308            0.07692
## Detection Prevalence     0.0000      0.0000     0.3077            0.69231
## Balanced Accuracy        0.5000      0.5000     0.9500            0.66667
##                      Class: MEDIUM LOW
## Sensitivity                     0.0000
## Specificity                     1.0000
## Pos Pred Value                     NaN
## Neg Pred Value                  0.7692
## Prevalence                      0.2308
## Detection Rate                  0.0000
## Detection Prevalence            0.0000
## Balanced Accuracy               0.5000

ave_predicted_as_ave <- confusion_matrix$table["AVE", "AVE"]
high_predicted_as_high <- confusion_matrix$table["HIGH", "HIGH"]
low_predicted_as_low <- confusion_matrix$table["LOW", "LOW"]
medium_high_predicted_as_medium_high <- confusion_matrix$table["MEDIUM HIGH", "MEDIUM HIGH"]
medium_low_predicted_as_medium_low <- confusion_matrix$table["MEDIUM LOW", "MEDIUM LOW"]

#How many actual "AVE" from the test data got predicted as "AVE"?
#=0

#How many actual "HIGH" from the test data got predicted as "HIGH"?
#=0

#How many actual "LOW" from the test data got predicted as "LOW"?
#=3

#How many actual "MEDIUM HIGH" from the test data got predicted as "MEDIUM HIGH"?
#=1

#How many actual "MEDIUM LOW" from the test data got predicted as "MEDIUM LOW"?
#=0

#The Classification went poorly since the model only has an accuracy of 30.77% which could be caused by the limited data collected.
#To make an improvement for the results, I would suggest on collecting more data to make the model more accurate and avoid false conclusions since the data only has 42 observations now.