htest

usage

Geting started

The programming language is python, please install python environment before using it,
conda with python 3.9+ is preferred, but other python envs are also fine.

Firstly You could create an env with the command like below or using your own env:

conda create --name=htest python=3.9

after created it successfully, you could activate it with this command:

conda activate htest

and then install the necessary packages(except for pytorch) with the command: pip install [package name]:

pandas  
scikit-learn  
torch_optimizer  
tensorboard

for pytorch installation, please refer to:
https://pytorch.org/get-started/locally/

If there are still other modules need to be installed or other problems, please install them accordingly(follow the hints), or open an issue, thank you.

How to run

I didn't upload the excel file because of privacy, but for simply testing, please rename the test excel file into "MealAnalysis(2017).xlsx", and put into the same folder with this file.

for machine learning approach

run with this command:
python ml.py
you will see the evaluation accuracy regarding the test data.

by the way, if with this line:
https://github.com/anguoyang/htest/blob/main/ml.py#L10
The accuracy will be 1.0 which means 100%, if comment this line, then accuracy is around 80%

for neural network approach

run with this command:
python nn.py
you will see the training status as well as the evaluation accuracy regarding the test data.

by the way, if with this line:
https://github.com/anguoyang/htest/blob/main/nn.py#L16
The accuracy around 1.0 which means 100%, if comment this line, then accuracy is around 80%

if you want to check the training history, please use this command:
tensorboard --logdir "./runs" and open your browser and input this URL:
http://localhost:6006/
you will see your training loss history and accuracy evaluation history.

***Please note that we decrease the target with 1 to let it start from 0 [0,3], so in real senarios, please add it back to the output to make sure the final output range [1,4] ***