Explain the project. Explain housing prices. Explain dataset features.
Explain linear regression. Explain math if time.
In order to
In my first experiment, I used all of the numeric features from the dataset. Any rows with null values were dropped for this experiment, about 300 rows. I wanted to see how well using all features predicted the data.
The coefficient of determination, the [explain], for this model was 0.799. [explain what that means.] The mean squared error, the [explain], for this model was 1,588,229,760.73. [what is this]. The root mean squared error was 39,852.60. This represents that the average distance between a data point's estimated value based on the model and the actual value is 39,852.60. [explain context]
The second experiment used most of the same features as the first experiment. However, instead of dropping rows with missing values, I dropped the two columns with the largest number of missing values: MasVnrType and BsmtExposure. I also deleted the three rows with missing values in the column MasVnrArea. I chose to try this to see if having more records would be more beneficial than having the additional features.
The coefficient of determination for this model was 0.848. The mean squared error was 930,983,000.66 and the root mean squared error was 30,512.02. [talk about hwo this is a lot better]