michaelistrofficus / hampel_filter Goto Github PK

Hello
I am experiencing an error while importing hampel lib with my python3.8 :
I am not sure what to do with this. Would you have some hints?
thank you in advance,
best regards
Yoann

from hampel import hampel
Traceback (most recent call last):
File "", line 1, in
File "/usr/local/lib/python3.8/dist-packages/hampel/init.py", line 1, in
from .hampel import hampel
File "/usr/local/lib/python3.8/dist-packages/hampel/hampel.py", line 5, in
from hampel.extension.hampel import hampel as hampel_extension
File "src/hampel/extension/hampel.pyx", line 1, in init hampel.extension.hampel
ImportError: numpy.core.multiarray failed to import (auto-generated because you didn't call 'numpy.import_array()' after cimporting numpy; use 'numpy._import_array' to disable if you are certain you don't need it).

Is the Hempel filter suitable for both nonparametric and normal data?

Hi,

I installed hampel on my computer and then on a server. It's called by a single python script using import hampel
when I run it manually or on a notebook it works perfect, however when it passes through a crontab I get this execution message

Traceback (most recent call last):
  File "/mellisphera/rundata.py", line 13, in <module>
    import hampel
ImportError: No module named 'hampel'

the server works with anaconda distribution
the module seems properly installed in
/opt/mellisphera/anaconda3/lib/python3.7/site-packages

and I've given it wide access rights to avoid any issue on that side

drwxrwxrwx  2 lorenzo mellisphera    4096 Jun  7 06:14 hampel-0.0.5.dist-info
-rwxrwxrwx  1 lorenzo mellisphera    2151 Jun  7 06:14 hampel.py

Any idea of why the batch mode cannot resolve this module (it resolves all others)?

Hi,
I am experiencing an "error" when applying hampel lib filter to data from a dataframe. Output data seems to have a bias in respect the original input data. Last lines of the output df are filled with NaN.
I am applying the filter as follows:

df_posthampel['A'] = hampel(df_prehampel['A'], window_size=10, n_sigma=3.0).filtered_data

Here a plot of the input raw data and the filtered:

Do you have news of something similar? Maybe I should modify window sizes, or just synchronize both columns after the filter.

Thank you in advance,
Best regards.

raise ValueError("Timeserie object must be of type pandas.Series.")

The window size should be 2*window_size + 1

Recent versions of pandas present a deprecation warning for Series.fillna, which is used in in this module:

 /usr/local/lib/python3.11/site-packages/hampel.py:49: FutureWarning:
Series.fillna with 'method' is deprecated and will raise in a future version. Use obj.ffill() or obj.bfill() instead.
/usr/local/lib/python3.11/site-packages/hampel.py:50: FutureWarning:
Series.fillna with 'method' is deprecated and will raise in a future version. Use obj.ffill() or obj.bfill() instead.

Indeed the rolling median computation uses the fillna().

The warning should be easily fixed by replacing the call by ffill() / bfill().

Hi,

great package! Is this maintained? Could you remove the fixed requirements?

        "numpy>=1.19.2",
        "pandas>=1.1.2"
    ],
    extras_requires={
        "dev": [
            "pytest>=6.0.2"
        ]
    },
    python_requires='>=3.7',

I guess for the few lines of code, it should be fine without them? Or at least could you lower them? Also the Python version would be great to make at least >=3.6