multyclass_logistic_regression

Realization own model of multyclass logistic regression

Aim of task:

1. Preprocessing data
2. Vizualiztion of data
3. Creation of model

Mine rule was not to use difficult function, write it myself.

What I realized myself:

1. Function "describe"
2. Vizualization of histogram
3. Scatter plot
4. Pair plot

Initially I had 2 datasets train_dataset and test_dataset. The first was used for training, the second - for prediction. Train_dataset has 4 classes and some features. Using Scatter Plot I need to choose 2 features for creation of model. In this project I used 2 methods of optimization - gradient descent and stochastic gradient descent and method for prediction - onevsall. Main goal was to get accuracy>=98%.