satishsolanki1990 / California_Housing_Price_LinearRegression

Linear Regression analysis of California housing price 1990 kaggle dataset.

California Housing Price (1990)

Introduction:

A linear regression analysis is performed on california houshing price(1990). Data were collected from kaggle. The motivation of this analysis is:

1. to study the impact of various Normalization technique.
2. First order Ablation study
3. Can we use weights of particular feature as a parameter for importance of that feature.
4. Feature selection techniques

Data Summary:

Raw Dataset consist of total 10 columns and 20640 examples.

Target feature = median_house_value

Input features = ['longitude', 'latitude', 'housing_median_age', 'total_rooms', 'total_bedrooms', 'population', 'households', 'median_income', 'median_house_value', 'ocean_proximity']

Appraoch:

1. Pre-processing

1.1 : Remove outliers

1.2 : Variance Inflation Factor (VIF) analysis

1.3 : Pairwise Correlation analysis using Heatmap

2. Experiments

2.1 : Without Normalization

2.2 : Standardization of input features

2.3 : Max-Min Normalization of input features

2.4 : log transformation of target feature + without Normalization

2.5 : log transformation of target feature + with Standardization

2.6 : First Order Ablation study

2.7 : Weight as parameter for Importance of features?

3. Post-processing: