https://github.com/flobz/bankruptcy-prediction
Bankruptcy prediction
https://github.com/flobz/bankruptcy-prediction
Last synced: about 1 month ago
JSON representation
Bankruptcy prediction
- Host: GitHub
- URL: https://github.com/flobz/bankruptcy-prediction
- Owner: flobz
- Created: 2021-01-10T18:49:32.000Z (over 5 years ago)
- Default Branch: master
- Last Pushed: 2021-01-10T22:33:28.000Z (over 5 years ago)
- Last Synced: 2025-11-18T14:35:53.016Z (8 months ago)
- Language: Jupyter Notebook
- Size: 9.06 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Bankruptcy prediction
I use a dataset that can be found [here](https://archive.ics.uci.edu/ml/datasets/Polish+companies+bankruptcy+data#).
The goal is to predict if a company will go bankrupt in the next 5 years.
You can see my work [here](https://github.com/flobz/Bankruptcy-prediction/blob/master/bankruptcy_prediction.ipynb)
# Context
This work was done for an exam, goals were:
- Understand a dataset
- Show interesting data with table and graph
- Use the best machine learning algorithm
- Optimize the algorithm
- Predict result
- Create an api to work with the model
# API
## Usage
You can test the model with the API that I developed. To use it you will need to install the requirement:
```pip install -r requirements.txt```
Then start the api:
```python3 api.py```
Then you need to create a request with data in json format like this:
```json
{ "nbyear": "a number between 1 and 5",
"data": ["64 financial data of the company"]
}
```
To create the "data" field see all attributes you need to specify [here](https://github.com/flobz/Bankruptcy-prediction/blob/master/data/attribute.txt).
You need to use the same order.
## Example
```commandline
curl -X POST -H "Content-Type: application/json" http://127.0.0.1:5000/is_bankrupt -d'{"nbyear": 1, "data": [0.20055, 0.37951, 0.39641, 2.0472, 32.351, 0.38825, 0.24976, 1.3305, 1.1389, 0.50494, 0.24976, 0.6598, 0.1666, 0.24976, 497.42, 0.73378, 2.6349, 0.24976, 0.14942, 43.37, 1.2479, 0.21402, 0.11998, 0.47706, 0.50494, 0.60411, 1.4582, 1.7615, 5.9443, 0.11788, 0.14942, 94.14, 3.8772, 0.56393, 0.21402, 1.741, 593.27, 0.50591, 0.12804, 0.66295, 0.051402, 0.12804, 114.42, 71.05, 1.0097, 1.5225, 49.394, 0.1853, 0.11085, 2.042, 0.37854, 0.25792, 2.2437, 2.248, 348690.0, 0.12196, 0.39718, 0.87804, 0.001924, 8.416, 5.1372, 82.658, 4.4158, 7.4277]}'
{
"Bankrupt": false,
"Information": "The company will not go bankrupt in 1 year."
}
```
```commandline
curl -X POST -H "Content-Type: application/json" http://127.0.0.1:5000/is_bankrupt -d'{"nbyear": 3, "data": [0.080622, 1.0208, 0.13118, 1.1542, -9.2273, -0.24848, 0.080622, -0.02034, 2.3527, -0.020763, 0.082926, 0.094766, 0.037078, 0.080622, 4271.1, 0.085457, 0.97961, 0.080622, 0.034267, 28.227, 1.6757, 0.082926, 0.034267, -0.16781, -0.16791, 0.085457, 35.993, 7.2508, 3.6773, 0.43301, 0.034805, 136.8, 2.6708, 2.2259, 0.082705, 2.3553, 58.736, -0.007143, 0.035153, 0.010993, 0.38002, 0.035247, 150.88, 122.66, 0.44311, 0.22486, 29.256, 0.076313, 0.032436, 0.9619, 0.85075, 0.37479, -1.1477, -0.39484, 624.0, 0.035204, -3.883, 0.96474, -0.65597, 12.931, 2.9758, 131.98, 2.7655, 130.05, 1.0]}'
{
"Bankrupt": true,
"Information": "The company will probably go bankrupt in 3 year."
}
```
## Conclusion
With this dataset we have a very good accuracy. I had only 21 false negative and no false positive on 1700 row.
The XGbost algorithm worked very well.