keyunluo / python-ffm Goto Github PK

import ffm
Traceback (most recent call last):
File "", line 1, in
File "/mnt/hgfs/VM_WorkFolder/software/python-ffm/ffm/init.py", line 1, in
from .ffm import FFMData, FFM, read_model
File "/mnt/hgfs/VM_WorkFolder/software/python-ffm/ffm/ffm.py", line 5, in
libffm_file = [i for i in os.listdir(path) if i.startswith('libffm.')][0]
IndexError: list index out of range

[root@HW2288 python-ffm]# /usr/bin/python2.7
Python 2.7.14 | packaged by conda-forge | (default, Mar 30 2018, 18:16:04)
[GCC 4.8.2 20140120 (Red Hat 4.8.2-15)] on linux2
Type "help", "copyright", "credits" or "license" for more information.

import ffm

return

No libffm.so Build!

my OS is Centos 7

Traceback (most recent call last):
File "sklearn-example.py", line 7, in
from ffm import FFM, FFMFormatPandas, FFMData, save_data
File "/media/lx/项目/tx/python-ffm/ffm/init.py", line 1, in
from .ffm import *
File "/media/lx/项目/tx/python-ffm/ffm/ffm.py", line 5, in
libffm_file = [i for i in os.listdir(path) if i.startswith('libffm.')][0]
IndexError: list index out of range

类FFMFormatPandas 中的transform方法中的
建议把 for _, row in tqdm(_df.iterrows(), total=len(_df)): 修改成

 for row in tqdm(df.itertuples(),total= len(df)):
            feature_tuple = []
            for cat in self.categorical:
                if pd.notnull(getattr(row, cat)):
                    feature_name = '{}_{}'.format(cat,getattr(row, cat))
                    ffm_value = (self.field_index[cat],self.feature_index[feature_name],1)
                    feature_tuple.append(ffm_value)

如果df 全部是数值型，在fit 的时候如果某个类别型的特征是int 型（如0），在transform的时候，通过 row[cat]来获取值的时候，会把int 型 转成float64类型（0->0.0)，导致在feature_index中找不到key。
另外itertuples()应该比iterrows()快。

IndexError Traceback (most recent call last)
in ()
----> 1 import ffm.ffm as ffm

~/比赛/Tencent/ffm/ffm/ffm.py in ()
3 import os
4 path = os.path.dirname(os.path.abspath(file))
----> 5 libffm_file = [i for i in os.listdir(path) if i.startswith('libffm.')][0]
6 lib_path = path + '/' + libffm_file
7

IndexError: list index out of range

score predicted by this package is different from libffm, by the way the model used is trained by libffm.

CentOS
in ffm.py
line:301 y_true = val_data.labels
=> y_true = [int(i) for i in val_data.labels]
on my machine print(set(y_true)) shows ['0','1'] while numpy could not recognize it as binary
[0,1] works

and add type cast in wrap_dataset would help reading data from texts :)
line: 105 node.f = int(f)
line: 106 node.j = int(j)
line: 107 node.v = float(v)
line: 119 d.label = float(y)

Hi,
I some questions about installing,hoping you help me,
when I use command:"python3 setup.py install "
it shows as follows:
running install
running bdist_egg
running egg_info
writing ffm.egg-info/PKG-INFO
writing dependency_links to ffm.egg-info/dependency_links.txt
writing requirements to ffm.egg-info/requires.txt
writing top-level names to ffm.egg-info/top_level.txt
reading manifest file 'ffm.egg-info/SOURCES.txt'
writing manifest file 'ffm.egg-info/SOURCES.txt'
installing library code to build/bdist.linux-x86_64/egg
running install_lib
running build_py
warning: build_py: byte-compiling is disabled, skipping.

running build_ext
building 'ffm.libffm' extension
gcc -pthread -B /home/guo/anaconda3/compiler_compat -Wl,--sysroot=/ -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -fPIC -Ilibffm -I/home/guo/anaconda3/include/python3.6m -c ffm/ffm-wrapper.cpp -o build/temp.linux-x86_64-3.6/ffm/ffm-wrapper.o -Wall -O3 -std=c++11 -march=native -DUSESSE -DDEBUG=0 -DLINUX
cc1plus: warning: command line option ‘-Wstrict-prototypes’ is valid for C/ObjC but not for C++
In file included from ffm/ffm-wrapper.cpp:48:0:
ffm/ffm-wrapper.h:8:17: fatal error: ffm.h: No such file or directory
#include "ffm.h"
^
compilation terminated.
error: command 'gcc' failed with exit status 1

should I change code in ffm.h?