Predicting House Sale Price (m240)

Hi everyone!

I’m posting my next project and waiting for your comments
and criticism from your side.

My two comments on the general content of the courses. I
restrained me and kept silent after the Statistical Hypotheses
module, I thought there are no worse modules, but after the Linear
regression module I want to say that I have not met a worse module
where you are thrown into the rapid flow of a mountain river with
cold water like a newly born puppy.

My comments on the general content of the courses.

  1. The Statistical hypotheses module is an obstinacy in p-value,
    about which there has been a fierce dispute for more than half
    a century. I had to spend a lot of time sorting out the essence of
    the issue - no references to any serious literature. Stop making
    links to Wikipedia that idiots write for idiots. No references to
    alternatives as Bayesian statistics and non-parametric statistics.
    Almost all statistics do not have an ideal distribution - why then do
    you give the p-value as true in the last instance.

  2. As for the Linear regression module, I can say that your
    the platform has a serious competitor in the face Jason Brownlee,
    since everything is available for free, using practical examples,
    step by step with a link to really good books, and not garbage that
    is printed only to cut money from popular topics. The cost of his
    books is comparable to the cost of your training, but at least it is
    clear what comes from where. The ML is a complex theme, do not forget
    that the level of education is falling in last time for example see
    SAT New York High School.

If you earn money - so give in exchange good content and not
meaningless articles in the stile “it is better to be beautiful and
healthy than poor and sick” what and how to learn step by step and
not a meaningless set of phrases like on the videos of some Udemy
courses with a strong accent
Predicting_House_Sale_Prices.ipynb (2.0 MB)
. I hope this information will be
useful to the platform management.

Should I open the case in case of
errors with calculations b(slope) and c(intersect) by Ridge
regression with StandardScaler, or will one of your employees find an
error in my code?
Best Regards, Vadim Maklakov

Click here to view the jupyter notebook file in a new tab

1 Like

Hi Vadim,

Thanks for sharing another cool project with the Community! As usual, very detailed and profound data analysis, great approach to apply both linear and ridge regression and compare the results, clean and easy-to follow code and appropriate use of code comments. This time I also liked how you formatted outputs from the code cells and also your variable naming is efficient.

About calculating b(slope) and c(intersect) - good question! From my point of view I cannot see any error in your code, so I’d suggest you to open a ticket in Scikit-learn. Or maybe further reviewres will find some issues, but probably it’s better not to wait for them and create a ticket.

Some suggestions from my side:

  • I noticed that you didn’t create any functions. Since you have some long and repeated blocks in your code (such as defining optimal alphas, creating scatter plots, etc.), introducing corresponding functions would help you to significantly optimize it.
  • Dataviz: consider making titles and labels bigger, sometimes they are difficult to read.
  • Avoid too long and technical subheadings (such as Train and test using LinearRegression with Ordinary Least Squares and OneHotEncoder and RobustScaler encoder). You can always add explanations in markdown afterward.
  • You can import several modules in one line. For example, instead of these lines:
from sklearn.linear_model import Ridge
from sklearn.linear_model import RidgeCV

use this:

from sklearn.linear_model import Ridge, RidgeCV
  • Remove the commented out code from the code cell [1].
  • It’s better to add an empty line before a code comment to improve code readability (the code cells [4], [11], etc.)
  • To create a list of N zeros, use this piece of code: [0]*N

Hope my suggestions were helpful. Keep up this fast learning pace and high quality of projects! :star2:

1 Like

Elena, thank you very much!
I keep in mind your notices.

1 Like