ibunny01 / mlnd

Udacity Machine Learning Curriculum

MLND: UDACITY MACHINE LEARNING

MATERIALS

MLND Lectures Materials Intro to Machine Learning(UD120)

ASSIGNMENTS

Titanic Survival Exploration

• A simple linear regression analysis is carried out prior to the full-scale machine learning process.
• Technologies used: Numpy, Pandas, Linear Regression

Assignments Result Github

Boston Housing

• Build a model to predict prices based on real estate data in the Boston area.
• DecisionTreeRegressor is used to learn data and GridSearch is used to optimize the algorithm.
• Technologies used: Numpy, Pandas, R2 score, Cross Validation, DecisionTreeRegressor, GridSearchCV

Assignments Result Github

Finding donors

• Based on the analysis that customers with income over 50K make donations, we leased other customers’ data to label more than 50K customers.
• Used various algorithms to analyze the results and how to optimize the algorithm.
• Technologies used: GaussianNaiveBayes, KNN, Logistic Regression, AdaBoost, F beta / Accuracy score

Assignments Result Github

Creating Customer Segmentation

• Visualize and analyze customer data from wholesale companies. Correlation of each feature point is obtained and correlation is checked
• Based on the insights obtained from the previous stage, PCA analyzes the feature points of each customer segment
• Technologies used: PCA scikit-learn, seaborn

Assignments Result Github

SmartCab

• Learn the model of taxi drivers using reinforcement learning.
• Proceeds learning based on limited environment and rules and optimizes performance using it.
• Technologies used: Numpy, Pandas, Q-Function

Assignments Results Github

Image Classification

Assignments Results Github

Digit Recognition

• Convolution Neural Network is used to project a project that recognizes five consecutive numbers. (Data using MNIST)
• Since there was no data set of consecutive images, the training set was constructed using the existing MNIST and SVHN datasets. And by learning this set, I was able to recognize a reasonable level of data.
• Technologies used: Tensorflow, Scipy, Numpy, Convolution Neural Network, Maxpooling, ReLu, L2 Regularizer

Assignments Results Github

SANDBOX

• Practice Code regardind Machine Learning

Practice Code for each machine learning algorithms

Sandbox for Machine Learning

Practice Code for Tensorflow

Sandbox for TensorFlow