Embedded Files
REAL ESTATE PRICE PREDICTION WITH MINIMAL DATA AND EXPERIMENTING TOP 3 REGRESSION MODELS.
Description
Description
IBM Data Science Capstone Project for Real Estate price prediction with minimal data.
IBM Data Science Capstone Project for Real Estate price prediction with minimal data.
This is a model made for prediction of Property with help of UCI Machine Learning Repository Datasets.
This model predicts the price of any property from its specific attributes
This is property price production for Boston, Massachusetts by using data as small as 550 rows. However, only 80% of it would be used for training the dataset reducing available data to only 440 rows. Hence, a dataset with 14 attributes including the price of the property was chosen.
By observing these matrices we can see the inter-correlation of the attributes.
On the right, we can get a broad view of the correlation of attributes with each other. We can see that price of a house is correlated to the number of immigrants in an area.
On the left, we can see clusterisation and the fashion of the graph. Here, despite immigrants being more correlated, the status of area has more exponential decay fashion hence easy predictability.
Also, I tried to experiment by adding an attribute which was made by basic Arithmetic operations of other attributes and we can observe the output has a great curve fashion. It also enhances the accuracy with +4.43%
THE FINAL MODEL GAVE AN AVERAGE ACCURACY OF ABOUT 84.1% AND HENCE WAS VERY SUCCESSFUL
Page updated
Google Sites
Report abuse