You can find the spreadsheets for this video and some additional materials here: drive.google.com/drive/folders/1sP40IW0p0w5IETCgo464uhDFfdyR6rh7 Please consider supporting NEDL on Patreon: www.patreon.com/NEDLeducation
Hi Divya, and glad you liked the video. As for your question, it depends on what hypothesis you want to test. If you are concerned with the individual relationships between variables, which is most often the case, t-stats are enough. F-tests are required only when you want to test for the significance of the model overall.
Hi Corrado, and glad you enjoyed the video! As I am have been moving through some portfolio optimisation topics as of lately (you might have seen videos on the Omega ratio and portfolio concentration already), will definitely touch upon the optimisation of VaR and Expected Shortfall as well. Thanks for the suggestion!
@@NEDLeducation one question: when the independent variables have different unit of measurement, do we need to standardize them before applying the LASSO?
@@corradoforza Hi Corrado, and thanks for the question! This is a common practice, you could z-score the variables by subtracting the mean and dividing by the standard deviation for additional robustness.
Excuse me, i want to ask sir. Why there is a different formula in objective function to minimize SSE, between sklearn library and this video. What is the different ?? Thx
@NEDL could you please elaborate why you have calculated the daily returns using the formula lower value/upper value-1, why this formula is used. Or in place of that, could we normalize the variables using log or anything? The results will be same or not?
Hi Aradhana, and thanks for the question! Yes, you can use log(lower value/upper value) in this case to get logarithmic returns. The results will be similar.
Hi Sheena, and thanks for the question! The "constant" column is needed if the matrix method for regression modelling is used (when coefficients and standard errors are obtained by multiplying a bunch of matrices together).