Module of 2nd year study about AI application on data analysis This repository contains projects and exercises completed by me for academic, self learning, and hobby purposes. Data analysis is presented in the form of iPython Notebooks.
- Decision Trees
Split dataset by decision tree which is based on entropy and information gain: write functions to calculate entropy and information gain, and functions to get the feature that maximizes information gain and use this feature to split the dataset.In the end, compare the Iterative Dichotomiser 3 and Sklearn decisiontreeclassifier. - Markov Decision Processs
Implement Markov Decision Process which provides a mathematical framework for modeling decision making, and it is useful for studying optimization problems by dynamic programming. In this practice, the MDP is based on utility, and in the end find the optimal policy for a classic environment wih 9 cells . - Probalistic Reasoning
Explore probalistic reasoning more about bayesian network by python.
-
Machine Learning in NetSecure NIDS
Implement the Network Intrusion Detection System (NIDS) based on Machine Learing algorithm given two datasets of network traffic：benign traffic and unknown traffic.Train the machine learning algorithm with benign traffic, with this trained NIDS, test the unknown traffic to determine whether the data is benign or normal.
The process of build trained NIDS:
1.Extract features.
2.Processing the data based on scaling techniques to separate features better.
3.Model/Parameter selection: this project uses One-Class SVM use RBF(Gaussian) kernel to define the geometric relationship between a feature vector X and support vectors Y.
4.Evaluation: test the unknown dataset after training NIDS with benigh dataset. -
Machine learning_ Kaggle Titanic survival prediction
Titanic dataset is from Kaggle, which is very famous for machine learning as a beginner. In this notebook,I focus more on exploring the data by using data visualization and how to select machine learning model for solving calssification and regression problems on supervised learning.This practice follow the process:
1.Data Import and Preprocessing
2.Explor Data
3.Feature Engineeing
4.Machine Learning Model Build 5.Feature Importance 6.Conclusion