How to run it?

[zccoder]

Thanks a lot for your work! I just want to get the format which is need for libffm.
So, I wonder What the input_file's format should be? and what is the output file's format?

[pklauke]

The input file is expected to be a simple csv file. Make sure the first column is the the target variable y and there are is no index or header in the csv file. If created with python pandas this could be created by
df.to_csv('train.csv', index=False, header=False)
or to also move the target (with column name 'target') to the first column:
df.loc[:, ['target', [c for c in df.columns if c != 'target']].to_csv('LibFFM/trainffm.csv', index=False, header=False)

That file can then be converted using LibffmGenerator.out trainffm.csv what will produce the format as described in https://github.com/guestwalk/libffm/blob/master/README.

Hope this helps.