In this code, I've used the glogistic regression technique from machine learning to estimate the curve of classification for the given dataset points. The data file Contains values for one variable, and so, the algorithm is implemented for single-variable datasets. Multi-variable algorithm will be coming soon.
The decision curve represents the probability of the variable being of one class or another (belongs or not, binary decision, computed as 0 or 1). As the value in the curve gets near 1, it means that the point has a higher probability of belonging to the classification.
The curve general form has the following structure:
f(x) = 1 / (1 + e^-(ax + b))
In the project, you will find 4 files. "logistic_regression.py" is the Python code to execute. There you will find the implementation of the algorithm and all the code. The other 3 files, whose names start with "log_reg_data..." are the datasets ready to test with the code. "log_reg_data.csv" contains data for logistic regression with two variables (linear multi-variable algorithm). It will be used for later implementations of the algorithm. The other files are different datasets, containing one variable and its classification. They are meant to test the program and traing it. You may use part of one file to train the system, and other to test it. Otherwise, you may want to test your own dataset with a particular feature. Go on and try it!
To run this code, you must have python installed. No other dependencies or libraries were used. You just have to download the dataset file and run the code. The console will output values of a and b for each step. Both files (code and dataset) must be in the same folder. Otherwise, you might have to change the path specified in en code qhere the dataset is imported.