SiweiMa

Ames-House-deep-learning

revisit Ames house prices project by using PyTorch

Ames-House-Prices-Multiple-Linear-Regression-Project-in-Python

Given Ames Housing dataset, the project started with an exploratory data analysis (EDA) to identify the missing values, suspicious data, and redundant variables. Then I performed a mixed stepwise selection to reduce the set of variables and select the best model based on AIC, BIC, and adjust R-squared. With the best model selected, the model assumptions were checked regarding normality, homoscedasticity, collinearity, and linearity between response and predictors. Several solutions were proposed to solve the assumption violation. The model was then tested on unseen data and scored on Root-Mean-Squared-Error (RMSE).

chatGFT

concrete-strength-prediction

Concrete is the single most widely used man-made material in the world. Construction workers rely on experiments to determine the strength of concrete. The app presents an attempt to predict the strength based on the information of raw materials by machine learning methods.

Human-Activity-Recognition

is-test

Implementation of an awesome trick to determine if training and test are from the same distribution

machine-learning-snippets

code snippet I created for machine learning workflow

network-analysis

police-stop-eda

The How, Who, When, and Where of Police Stops in San Francisco: a case study of visualization and EDA

SiweiMa.github.io

Siwei Ma's personal website

SiweiMa/Ames-House-Prices-Multiple-Linear-Regression-Project-in-Python

Given Ames Housing dataset, the project started with an exploratory data analysis (EDA) to identify the missing values, suspicious data, and redundant variables. Then I performed a mixed stepwise selection to reduce the set of variables and select the best model based on AIC, BIC, and adjust R-squared. With the best model selected, the model assumptions were checked regarding normality, homoscedasticity, collinearity, and linearity between response and predictors. Several solutions were proposed to solve the assumption violation. The model was then tested on unseen data and scored on Root-Mean-Squared-Error (RMSE).

SiweiMa/concrete-strength-prediction

Concrete is the single most widely used man-made material in the world. Construction workers rely on experiments to determine the strength of concrete. The app presents an attempt to predict the strength based on the information of raw materials by machine learning methods.

SiweiMa/Ames-House-deep-learning

revisit Ames house prices project by using PyTorch

SiweiMa/chatGFT

SiweiMa/Human-Activity-Recognition

SiweiMa/is-test

Implementation of an awesome trick to determine if training and test are from the same distribution

SiweiMa/machine-learning-snippets

code snippet I created for machine learning workflow

SiweiMa/network-analysis

SiweiMa/police-stop-eda

The How, Who, When, and Where of Police Stops in San Francisco: a case study of visualization and EDA

SiweiMa/SiweiMa.github.io

Siwei Ma's personal website

SiweiMa/snake-game

A classic snake game with some modifications.

SiweiMa/topic-modeling-jeopardy

SiweiMa/user-search

Given a customer ID, the tool will return the customer's segment/clustering, CLV, the conditional expected number of purchases for next month, the probability of the customer being alive, detailed purchase history.