FancyImpute installation in Anaconda

Question

I was trying to install fancyimpute library on my Windows system. I am using Spyder on the Anaconda Navigator. I have tried the following on the command prompt as Administrator, still it shows

ModuleNotFoundError: No module named 'fancyimpute'

conda install -c https://conda.binstar.org/eswears cvxopt
pip install keras
pip install fancyimpute
Also installed Visual C++ Compiler for Python 2.7
I am using Python 3.6.1

Kindly help :)

Solution 1

Thanks MyopicVisage. Your suggestion worked for me with a minor change as below.

conda install ecos  
conda install CVXcanon  
pip install fancyimpute  

Solution 2

This is how I solved this problem in my laptop. First install MS Build tool Visual Studio Studio Build Tools requires about 3 GB of disk space. After installation of visual studio installer, look for the required version of C++ compiler (in our case its 14.0.xxx) Install that version which will download a file of 900 MB and requires a space of 3GB around in the Hard disk. After installation, restart the computer. Then go to conda prompt and give these commands

conda update --all
conda create -n py36 python=3.6 anaconda
conda activate 
conda install ecos
conda install CVXcanon
conda install -c cvxgrp scs
conda install scs
pip install fancyimpute

Another reference to do this, Unable to install fancyimpute in Python (for Windows 10)

If there's any error regarding "wrapt" , just go through this https://github.com/tensorflow/tensorflow/issues/30191 After this you might need to reinstall sklearn and jiblib lilbraries again

pip  uninstall sklearn
pip uninstall jiblib
pip  install sklearn
pip install jiblib

Solution 3

I found installing the dependent libraries with conda would allow you to install fancyimpute. Specifically, I had to install ecos and CVXcanon; you may need to install others, do so in the order of the warning messages.

Use the commands in the command prompt:

conda install ecos  

conda install CVXcanon  

conda install fancyimpute  

Solution 4

you can do pip install fancyimpute-0.0.4.tar.gz to install the package after downloading fancyimpute-0.0.4.tar.gz.

Solution 5

For those of you that couldn't get it working with other solutions, here is what I did. First:

conda install ecos

Then I went and downloaded the build wheel for both scs and cvxpy.

To install a build wheel, all I did was navigate to the folder (in my case, in the anaconda prompt it was just cd downloads). Secondly, I ran the following commands, in order:

pip install scs-2.1.0-cp36-cp36m-win_amd64.whl
pip install cvxpy-1.0.22-cp36-cp36m-win_amd64.whl

Then at last I could do

pip install fancyimpute

Side Note: that you can also copy paste the path to the file, pip just needs to know where it is:

pip install C:\Users\<USERNAME>\Downloads\scs-2.1.0-cp36-cp36m-win_amd64.whl
pip install C:\Users\<USERNAME>\Downloads\cvxpy-1.0.22-cp36-cp36m-win_amd64.whl

View more solutions

Comments

[Quickbeam2k1]

A different idea could be to use docker to construct exactly the python environment you want. Specifically I had some problem with airflow and xgboost. But with docker it was quite easy to run both.

[LR_RL]

For those Who try MyopicVisage's solution and run into this error: cannot run rc.exe, This is the solution: stackoverflow.com/questions/35215971/…

[AMC]

Does this answer your question? PackagesNotFoundError: The following packages are not available from current channels:

[Astrid]

This worked for me on Ubuntu 18.04 with Anaconda with py.3.6.9.

[AMC]

Isn't easy_install fancyimpute supposed to replace pip install fancyimpute ?

[Gian Arauz]

Hi AMC. When I gave this answer, easy_install worked fine. I don't know if pip install should be preferable than easy_install, I'm sorry. This easy_install solution was originally proposed by Nick in this link. I'm sure he has more knowledge that me and he can help you.