I would prefer not using the word "from scratch" in the title of the videos as this video is not for newbies to the ML who want to code linear regression model and you are using build in one....
i am confused. why do we want to do complicated things when we can do simple things please send me your thoughts. this video is great with all the explanations
Awesome, I am having a challenge, after running the last part of the code (sklearn_model = LinearRegression().fit(X_train,y_train) sklearn_y_prediction = sklearn_model.predict(X_train) it keeps on giiving me this error () sklearn_y_prediction.shape) (Input contains NaN, infinity or a value too large for dtype('float64'). I have checked to see if the are any NaN values but there are none
if i have a sample (e.g., ftse100) and I have 10 years worth of data for each of my variables: do i need to do 10 separate regressions (for each year); or, take average (of yearly data points) over 10 years and do a single one. OR, could I somehow do a single regression whilst including all data points ??
Here, we manually checked the correlation and eliminated those that are not significant. Is there any way to implement an algorithm that does it for us?
@@GregHogg Your video is misleading.. From scratch means building regressor object which can do normalization, gradient descent, predict values from answer, accuracy visualization etc etc using Numpy and Matplotlib. EQUATIONS!!!!
Take my courses at mlnow.ai/!
I would prefer not using the word "from scratch" in the title of the videos as this video is not for newbies to the ML who want to code linear regression model and you are using build in one....
indeed
@GregHogg Thank you Sir. You have really helped me to do my Exam today. More Blessing Sir
i am confused. why do we want to do complicated things when we can do simple things please send me your thoughts. this video is great with all the explanations
Awesome, I am having a challenge, after running the last part of the code (sklearn_model = LinearRegression().fit(X_train,y_train)
sklearn_y_prediction = sklearn_model.predict(X_train)
it keeps on giiving me this error ()
sklearn_y_prediction.shape)
(Input contains NaN, infinity or a value too large for dtype('float64').
I have checked to see if the are any NaN values but there are none
Have you made sure of this by removing all of the null rows in the data frame with df = df.dropna()?
@@GregHogg Yes just did and it works fine now. thanks.
@@aondonguaddai1207 Awesome
If you want to be a Data Scientist in Python, forget it!! R is the tool for REAL data science/analytics!
Thank you so much for this video
Very welcome!
Awesome video... thanks👍
You're very welcome!!
Awesome!!!
Woot!!!
if i have a sample (e.g., ftse100) and I have 10 years worth of data for each of my variables:
do i need to do 10 separate regressions (for each year); or, take average (of yearly data points) over 10 years and do a single one.
OR, could I somehow do a single regression whilst including all data points ??
from where can i access the dataset
Here, we manually checked the correlation and eliminated those that are not significant. Is there any way to implement an algorithm that does it for us?
Yeah there's an algorithm for everything really haha
what to do if the x-axis is time series ?
So is it true that f(x1, x2, x3) = p1x1 + p2x2 + p3x3 where pi is the ith value of the prediction model?
And how do you know what pi is? Thanks
brother this is not from scratch
Jump to 1:49 to skip bla bla bla ...
great video !!!
Thank you.
@@GregHogg Your video is misleading.. From scratch means building regressor object which can do normalization, gradient descent, predict values from answer, accuracy visualization etc etc using Numpy and Matplotlib. EQUATIONS!!!!
Bro ruined the whole video just by importing the model.
Looking back on this video, I completely 100% agree. I think I forgot in the middle that I was supposed to do that part in numpy
where is the dataframe?
Hi , Check on the spelling of dataframe. It should be DataFrame. Not Dataframe