The maxsnippetmodel's discuss from jostmey

maxsnippetmodel's Issues

Cannot improve accuracy even if even training data is very simple

Thank you for publishing this wonderful software.

I would like to use this for my data.

Firstly, when we ran the program on the whole data, the accuracy did not improve from 50%,
so I tried to prepare a simple demonstration data.

I uploaded the demo data in zip format.

demo_data.zip

That is, the structure of the directory when running the program is as follows.

├── bias.txt
├── data
│   ├── answers.csv
│   ├── sample_1_negative.csv
│   ├── sample_1_positive.csv
│   ├── sample_2_negative.csv
│   └── sample_2_positive.csv
├── dataplumbing.py
├── dataplumbing.pyc
├── dataplumbing_synthetic_data.py
├── lib
│   ├── aminoacid_factors.py
│   ├── atchley_factors.csv
│   ├── atchley_factors.py
│   ├── atchley_factors.pyc
│   ├── dna_factors.py
│   └── __pycache__
│       └── atchley_factors.cpython-36.pyc
├── lib_paths.py
├── lib_paths.pyc
├── LICENSE
├── model.py
├── model.pyc
├── __pycache__
│   ├── dataplumbing.cpython-36.pyc
│   ├── lib_paths.cpython-36.pyc
│   └── model.cpython-36.pyc
├── README.md
├── score.py
├── train.py
└── weights.txt

And then, run the command below.