Housing Prices Prediction

in this python code I used House Prices dataset from Kaggle website to train a linear regression model and try to predict the price of the house based on given features.

Dealing with Data:

• Data Importing

Using pandas inbuilt function pd.read_csv( ) I imported data from the given CSV file into my code.

• Filling missing values

For numerical values, I used KNeighborsRegressor to impute for missing numerical values based on other given values in the same column.

For categorical values, Some of the missing values are not actually missing but N/A means something from the data descrption file so I replaced the missing values for these features to None. And for the other features that N/A doesn't mean something I used mode imputation method to impute for those missing values.

• transforming Categorical values to string

from "data_description.txt" I found some featurs with type "object" that have numerical values like "MSSubClass" feature so in order not to let the model deal with it as numerical value I converted its data to string to make sure it won't take unwanted values that doesn't represent any thing

• Reducing Skewing of Data

I used log transformation method to reduce the skew of data and make it more closer to normal distribution.

• Replace categorical data (strings) with numerical values

I replaced data with categorical datatype and gave them unique values.

• Adding Bias Column

• Splitting the dataset

Finally, I splitted the dataset into training data to train the model on and validation data to test the model and check for its accuracy

Linear Regression Model:

I created a class for the linear regression model and used the Normal Equations to train the model and calculate the weights then used these weights to predict the output for new values.

Calculating the Error

I used MAE (mean absolute error) method to check for the model accuracy and calculate the error between the predicted and the actual values

Comparing with Linear Regression model from sklearn

In the Last step I compared between the model I created and the built-in model from sklearn linear_model library to make sure it will make the same predictions.

That's All. I hope you found this useful. Thank you for reading :).