/CS241-Problem-Solving-and-Practice-Project

Primary LanguageC++

Predict US Adult Income Using Logistical Regression

Brief Introduction

This is the course project for Problem Solving and Practice Course(CS241), which solves a classification task using logistical regression.
The data set is 'US Adult Income' on Kaggle ('https://www.kaggle.com/johnolafenwa/us-census-data'), containing 48842 records.

Data Preprocessing

1.Define and remove dirty data.
2.Transfer some data into one-hot encoding.
3.Normalize some data entries.

Data Visualization

Design a small tool using FLTK which can generate bar charts, line charts, and so on.

Logistical Regression

Complete a logistical regression algorithm with only C++ STL, and reach accuracy 78.43%, better than DummyClassifiers in Pythom. Here shows the pseudo code:

while iteration_time > 0
     Y'=sigmoid(X*W);   
     for i in Y':  
     Loss=-Y[i]*log(Y'[i])-(1-Y[i])*log(1-Y'[i]); //count loss  
     dW=X^T*(Y'-Y);  
     W=W-dW*learning_rate;  //updata W using gradient descent  
     iteration_time--;  
end While