Determining the sales price of a house based on suitable predictor variables (24) of 71 variables, best describing the house to predict the sales price
2. Various factors affect the Sale Price of houses , in this problem we are tasked to
predict the Sales Price of a house based on various factors or attributes.
The Data Set is a large data set with 79 explanatory variables describing almost every
aspect of residential homes in Iowa .
Problem Scenario
3. Correlation Matrix
Correlation of 24 Predictor
variables (various factors of a
house ) with Dependent
variable ( House Sale’s Price )
Plotted correlation with
condition (correlation greater
than .325 or less then -.325)
These 24 variables satisfy the
above condition and are
chosen variables
4. <<<<<-Performing Data Wrangling
This code takes in data set from .csv file & reads all the data
Dimentions: ( 1460 * 71)
This code then converts categorical data into numeric
Continuation ->>>>>
Post conversion the converted data is written into a
separate file.
NA or corrupt values are removed & correlation of all
predictor variables against dependent variable ( Sales
Price ) needs to be produced
5.
6. Results
The R Square value for the model to
predict the House Sale’s Price is: 80.02%
F-Statistics value is 183.4 , which is high
and much significant
Sale's Price Of The House = -338801.122 + 44.641*(LotFrontage) + 14350*(OverallQual) + 147.29*(YearBuilt) + 204.426*(YearRemodAdd) +
24.821*(MasVnrArea) -8423.37*(ExterQual) - 512.736*(Foundation) - 8343.79*(BsmtQual) + 18.21*(BsmtFinSF1) + 2.212*(TotalBsmtSF) - 1156.93*(HeatingQC)
+6.539*(X1stFlrSF) + 33.956*(GrLivArea) - 1629.20*(FullBath) - 10201.50*(KitchenQual) +2387.47*(TotRmsAbvGrd) + 8123.94*(Fireplaces) -
412.45*(GarageType) - 146.46*(GarageYrBlt) - 793.66*(GarageFinish) + 13993.63*(GarageCars) + 7.60*(GarageArea) + 27.55*(WoodDeckSF)-
4.61*(OpenPorchSF)
7. Not Accurate, Though! , Predicted House Sales Price close to Actual Sales Price Of the House
8. Still Working On This Mode To Improve The Model Prediction Accuracy ………
Please Leave comments if you have some ideas on how to make the model
predict closer to actual values!!