Note - Before using the "test_stationarity" created in the video, please remove the "NaN" values first and then use this function for better result... e.g. df.dropna() test_stationarity(df, 'your variable') ==== you can also try df.dropna(inplace=True) sorry for the inconvenience
This is the best video I have found to understand how to make the data stationary. Especially, almost no video shown the mix of operations you did to the time series.
I'm getting different numbers for log + sqrt + shift, as well as the first shift. When you applied dropna after getting the error, it seems to have removed the rolling values.
Thanks for bringing it into the notice. Rolling statistics remains there, but what happens is, first it removes the na value in the supplied variable and then whatever is left, on top of that it calculates the rolling statistics.... but the rolling statistics remains unimpacted.. and on overall level also there won't be much impact as it's just removing one month before calculating the adfuller... in case if you are shifting for more than a month, recommend you to do the adfuller test separatly..... good observation....
U seem to have the issue that it is not showing all of the columns u might be getting ........ as well if so, run this pd.set_option('display.max_columns
hello, thanks for this great video, my problem is it doesn't plot the mean and standard deviation properly for my data, for example it show's mean of half of my data and also my result for logShiftDiff in test statistics indicate that data is stationary but the plot isn't correct by the way I write your codes for plotting and test_stationarity
test_stationarity(air_df.dropna(),'shiftdiff'), while checking stationarity with test_stationarity function am getting a value error i.e ValueError: Could not interpret value `shiftdiff` for parameter `y` can you help me to solve above problem
here you can find all of the data science ones - github.com/LEARNEREA/Data_Science/tree/main/Scripts for other you can just go on the home page of this GitHub link and can check out whatever you want from respective repository
Note - Before using the "test_stationarity" created in the video, please remove the "NaN" values first and then use this function for better result... e.g.
df.dropna()
test_stationarity(df, 'your variable')
==== you can also try df.dropna(inplace=True)
sorry for the inconvenience
This is the best video I have found to understand how to make the data stationary. Especially, almost no video shown the mix of operations you did to the time series.
Glad it was helpful
I'm getting different numbers for log + sqrt + shift, as well as the first shift. When you applied dropna after getting the error, it seems to have removed the rolling values.
Thanks for bringing it into the notice.
Rolling statistics remains there, but what happens is, first it removes the na value in the supplied variable and then whatever is left, on top of that it calculates the rolling statistics.... but the rolling statistics remains unimpacted.. and on overall level also there won't be much impact as it's just removing one month before calculating the adfuller... in case if you are shifting for more than a month, recommend you to do the adfuller test separatly.....
good observation....
In 7:15 i wrote df.head() and my data comes only with the columns: passenger, rollMean and rollStd. In yours there are more, is sth wrong ?
U seem to have the issue that it is not showing all of the columns u might be getting ........ as well if so, run this
pd.set_option('display.max_columns
hello, thanks for this great video, my problem is it doesn't plot the mean and standard deviation properly for my data, for example it show's mean of half of my data and also my result for logShiftDiff in test statistics indicate that data is stationary but the plot isn't correct
by the way I write your codes for plotting and test_stationarity
If possible, share you script at learnerea.edu@gmail.com
test_stationarity(air_df.dropna(),'shiftdiff'), while checking stationarity with test_stationarity function am getting a value error i.e ValueError: Could not interpret value `shiftdiff` for parameter `y` can you help me to solve above problem
Why do have written shiftdiff in quotes?
can you please share the python code explained in all the videos
here you can find all of the data science ones -
github.com/LEARNEREA/Data_Science/tree/main/Scripts
for other you can just go on the home page of this GitHub link and can check out whatever you want from respective repository
thank you broda
glad it was helpful
Github code link ?
Hi Sir your all videos are awesome but I did not find video oh hypertuning n PCA.
Will make it soon