Machine learning
Question 1
The optimal alpha value for the ridge regression was obtained to be 1000, and the optimal alpha value for the lasso regression was obtained to be 0.0001. When you change the alpha values, the coefficients of the independent variables and the model accuracy increases for both the lasso and ridge regression. The best predictors of the models are ScreenPorch, PoolArea and MiscVal.
Question 2
The ridge and lasso regression have different lambda values. To determine the best model, we will look at the accuracy of the model. The alpha value for the Lasso regression will be selected because it produces a higher regression value, i.e. it is more accurate
Question 3
After rebuilding the model after excluding the five most import variables, we obtained the five most important predictors to be GarageArea, WoodDeckSF, OpenPorchSF,TotalBsmtSF and 2ndFlrSF
Question
The make sure that the model is robust and generalized, one should work on extreme values. The extreme values can either be the outliers or the Novelties. These absolute values in the model cause wrong prediction and also false accuracy. This later leads to faulty decision making from the organizations[1]. The same applies to the unstandardized variables i.e. some variables have larger values while other have very low value[2]. The larger values needs to be standardized before making predictions. These cause inaccurate and they should be dealt with before creating the model.
Biblography
Didona, D., Quaglia, F., Romano, P., & Torre, E. (2015, January). Enhancing performance prediction robustness by combining analytical modeling and machine learning. In Proceedings of the 6th ACM/SPEC international conference on performance engineering (pp. 145-156). ACM.
Tsipras, D., Santurkar, S., Engstrom, L., Turner, A., & Madry, A. (2018). Robustness may be at odds with accuracy. arXiv preprint arXiv:1805.12152.
[1] Didona, D., Quaglia, F., Romano, P., & Torre, E. (2015, January). Enhancing performance prediction robustness by combining analytical modeling and machine learning. In Proceedings of the 6th ACM/SPEC international conference on performance engineering (pp. 145-156). ACM.
[2] Tsipras, D., Santurkar, S., Engstrom, L., Turner, A., & Madry, A. (2018). Robustness may be at odds with accuracy. arXiv preprint arXiv:1805.12152.