awslabs / datawig

Imputation of missing values in tables.
Apache License 2.0
478 stars 69 forks source link

Segment 11 warning #117

Closed shoraj closed 4 years ago

shoraj commented 4 years ago

Unable to run datawig code in google colab

felixbiessmann commented 4 years ago

Could you provide any hints on what errors you got?

shoraj commented 4 years ago

Whenever I am trying to run the "import datawig" command the google collab session is crashing and restarting. I tried to check the log it showed me a warning which says Segmentation fault: 11 and [bt] (0) /usr/local/lib/python3.6/dist-packages/mxnet/libmxnet.so(+0x3c27360) [0x7f813614c360] .

[image: Mailtrack] https://mailtrack.io?utm_source=gmail&utm_medium=signature&utm_campaign=signaturevirality5& Sender notified by Mailtrack https://mailtrack.io?utm_source=gmail&utm_medium=signature&utm_campaign=signaturevirality5& 03/13/20, 04:38:10 PM

On Fri, Mar 13, 2020 at 4:00 PM felixbiessmann notifications@github.com wrote:

Could you provide any hints on what errors you got?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/awslabs/datawig/issues/117#issuecomment-598655371, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACHSEIXUASQ5TXX2GEJ5MB3RHIDN7ANCNFSM4LG7OIWQ .

felixbiessmann commented 4 years ago

Hm, not sure to what extent this is a datawig issue, also we never have tested this in colab, but we'll look into it asap (probably after the ICML review deadline next week).

shoraj commented 4 years ago

Thank You.

[image: Mailtrack] https://mailtrack.io?utm_source=gmail&utm_medium=signature&utm_campaign=signaturevirality5& Sender notified by Mailtrack https://mailtrack.io?utm_source=gmail&utm_medium=signature&utm_campaign=signaturevirality5& 03/13/20, 04:47:05 PM

On Fri, Mar 13, 2020 at 4:44 PM felixbiessmann notifications@github.com wrote:

Hm, not sure to what extent this is a datawig issue, also we never have tested this in colab, but we'll look into it asap (probably after the ICML review deadline next week).

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/awslabs/datawig/issues/117#issuecomment-598671953, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACHSEITA5BMAYH5ENDOJKCLRHIIQVANCNFSM4LG7OIWQ .

felixbiessmann commented 4 years ago

so it seems to be a backwards incompatibility issue with mxnet, picking a random older version seems to work. the below example just ran through in a colab notebook:

!pip install mxnet==1.3.0
!pip install datawig

import datawig, numpy

# generate some data with simple nonlinear dependency
df = datawig.utils.generate_df_numeric() 
# mask 10% of the values
df_with_missing = df.mask(numpy.random.rand(*df.shape) > .9)

# impute missing values
df_with_missing_imputed = datawig.SimpleImputer.complete(df_with_missing)

df_with_missing_imputed

We'll fix the mxnet version number in the next release, sorry about that.

shoraj commented 4 years ago

Don't be sorry, actually, I had tried running datawig a few days ago, it was absolutely fine. only from yesterday, I am facing the problem. All the best, thank you for your immediate reply.

[image: Mailtrack] https://mailtrack.io?utm_source=gmail&utm_medium=signature&utm_campaign=signaturevirality5& Sender notified by Mailtrack https://mailtrack.io?utm_source=gmail&utm_medium=signature&utm_campaign=signaturevirality5& 03/13/20, 04:52:12 PM

On Fri, 13 Mar 2020, 4:50 pm felixbiessmann, notifications@github.com wrote:

so it seems to be a backwards incompatibility issue with mxnet, picking a random older version seems to work. the below example just ran through in a colab notebook:

!pip install mxnet==1.3.0!pip install datawig import datawig, numpy

generate some data with simple nonlinear dependency

df = datawig.utils.generate_df_numeric() # mask 10% of the values df_with_missing = df.mask(numpy.random.rand(*df.shape) > .9)

impute missing values

df_with_missing_imputed = datawig.SimpleImputer.complete(df_with_missing)

df_with_missing_imputed

We'll fix the mxnet version number in the next release, sorry about that.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/awslabs/datawig/issues/117#issuecomment-598674165, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACHSEIUWP7JCDRNAB5O4LO3RHIJJHANCNFSM4LG7OIWQ .

felixbiessmann commented 4 years ago

Should be solved with latest release, please reopen if problem persists