Can anyone help with Nan in R. Everything is fine with an exception in the calculation of RSME and MAPE. It kept giving me a NaN. I tried to filter out the zero columns but still an issue

Here is the info

The file BostonHousing.csv below contains information collected by the US Bureau of the Censusconcerning housing in the area of Boston, Massachusetts. You will use it to practice data mining in R.The dataset includes information on 506 census housing tracts in the Boston area.

The goal is to predict the median house price in new tracts based on information such as crime rate, pollution, and number of rooms. The dataset below contains 13 predictors, and the response is the median house price (MEDV).

*1) why should the data be partitioned into training and validation sets? What will the training set be used for? What will the validation set be used for?

*2) Fit a multiple linear regression model to the median house price (MEDV) as a function of CRIM, CHAS, and RM. Write the equation for predicting the median house price from the predictors in the model.

*3) Using the estimated regression model, what median house price is predicted for a tract in the Boston area that does not bound the Charles River, has a crime rate of 0.1, and where the average number of rooms per house is 6?

*4) Reduce the number of predictors:

* a) Which predictors are likely to be measuring the same thing among the 13 predictors? *Discuss the relationships among INDUS, NOX, and TAX.

* b) Compute the correlation table for the 12 numerical predictors and search for highly *correlated pairs. These have potential redundancy and can cause multicollinearity. *Choose which ones to remove based on the above table

c) Use stepwise regression with the three options (backward, forward, both) to reduce the remaining predictors as follows: Run stepwise on the training set. Choose the top model from each stepwise run. Then use each of these models separately to predict the validation set. Compare RMSE, MAPE, and mean error, as well as lift charts. Finally, describe the best model.

dataset- https://github.com/selva86/datasets/blob/master/BostonHousing.csv

Here is my codes

library(correlation)

library(ggplot2)

library(dplyr)

library(broom)

installed.packages("ggpubr")

installed.packages("Metrics")

installed.packages("PerformanceAnalytics")

installed.packages("DescTools")

library(DescTools)

read.csv(file.choose(BostonHousing))

housing.df<-data.frame(BostonHousing)

# (1) Tell me why should the data be partitioned into training and validation sets? What will the training setbe used for?

# What will the validation set be used for?

# We need to partition the data into the training and validation set for the purpose

# of assessing the generalization of the model; also, it is useful to identify and

# evaluate the relationships between the predictor and predicted variables.

# The training data is used for the purpose of model fitting. And, the validation data

# is used for empirical validation and measures of errors.

#(2) Fit a multiple linear regression model to the median house price (MEDV)

# as a function of CRIM,CHAS, and RM. Write the equation for

# predicting the median house price from the predictors in the model.

# We will split the data 70% for training, and 30% for validation

reg <-lm(MEDV~CRIM+CHAS+RM, data=housing.df)

summary(reg)

# The linear regression model is MEDV= -28.81 + (-.261*CRIM)

# + (3.76*CHAS) + (8.28*RM)

#(3) Using the estimated regression model, what median house price is predicted for a tract in the

# Boston area that does not bound the Charles River, has a crime rate of 0.1, and where the average number of rooms per house is 6?

# MEDV= -28.81 + (-.261*.1) + (3.76*0) + (8.28*6)

reg$coef%*%c(1,0.1,0,6)

# The median house price is $20,832.32

# (4)(a) Reduce the number of predictors: Which predictors are likely to be measuring the same thing among the 13 predictors?

# (b) Discuss the relationships among INDUS, NOX, and TAX.

# (a) Some of the predictors are likely to measure the same thing, but in different ways.

# Some of the predictors are : ZN, INDUS, Tax.

# All this provides a proportion related to the area of land, house.

indus=housing.df$INDUS

nox=housing.df$NOX

tax=housing.df$TAX

d=data.frame(indus,nox,tax)

cor(d)

# correlation between indus and nox is .7636; correlation between indus and tax is .7208, and

# correlation between nox and tax is .6680

# (b) There is a high correlation between INDUS, NOX, and TAX as they include

# a higher percentage of non-retail businesses that translate to higher pollution and taxes.

# (c) Compute the correlation table for the 12 numerical predictors

# and search for highly correlated pairs. These have potential redundancy and can cause multi-collinearity.

# Choose which ones to remove based on the above table

crim=housing.df$CRIM

zn=housing.df$ZN

chas=housing.df$CHAS

rm=housing.df$RM

age=housing.df$AGE

dis=housing.df$DIS

rad=housing.df$RAD

ptratio=housing.df$PTRATIO

lstat=housing.df$LSTAT

data=data.frame(crim,zn,indus,chas,nox,rm,age,dis,rad,tax,ptratio,lstat)

cor(data)

# There is a high positive correlation between nox and indus = 0.7637

# There is a high positive correlation between rad and tax = 0.91022

# There is a high negative correlation between dis and nox = -0.7692

# We might remove the nox predictor according to the given matrix

# (d) Use stepwise regression with the three options (backward, forward, both)

# to reduce the remaining predictors as follows: Run stepwise on the training set.

# Choose the top model from each stepwise run. Then use each of these models separately to predict the validation set.

# Compare RMSE,MAPE, and mean error, as well as lift charts. Finally, describe the best model.

# Stepwise regression

#The models with minimum AIC are:

# Backward: medv ~ crim + zn + chas + nox + rm + dis + rad + tax + ptratio +lstat

# Formard: medv ~ crim + zn + indus + chas + nox + rm + age + dis + rad + tax + ptratio + lstat

# Both: medv ~ crim + zn + chas + nox + rm + dis + rad + tax + ptratio +lstat

spec = c(train = .7, validate = .3)

df=data.frame(crim,zn,indus,chas,nox,rm,age,dis,rad,tax,ptratio,lstat,medv)

df=data.frame(housing.df)

g = sample(cut(

seq(nrow(df)),

nrow(df)*cumsum(c(0,spec)),

labels = names(spec)

))

res = split(df,g)

train=res$train

validate=res$validate

library(DescTools)

model1=lm(medv~ crim + zn + rm +age+ rad + tax + ptratio + lstat, data=validate)

summary(model1)

RMSE(validate$medv,model_backward$fitted.values)

MAPE(validate$medv,model_backward$fitted.values)

step(model1,direction = "backward")

model2=lm(medv~ crim + zn + chas + nox + rm + age + dis + rad + tax + ptratio + lstat,data=validate)

summary(model2)

step(model2, direction = "forward")

model3=lm(medv~crim + zn + chas +indus, nox + rm + dis + rad + tax + ptratio +lstat,data=validate)

summary(model_both)

step(model_both, direction = "both")

Here is my filtering out the NaN

housing.df

is.na(housing.df)

which (is.na.data.frame(housing.df))

unique (unlist (lapply (housing.df, function (x) which (is.na (x)))))

na.exclude(housing.df)

model_backward=lm(medv ~ crim + zn + chas + nox + rm + dis + rad + tax + ptratio + lstat,data=validate)

summary(model_backward)

RMSE(validate$medv,model_backward$fitted.values)

MAPEvalidate$medv,model_backward$fitted.values)