Machine-Learning-Project

These are my projects for machine learning. I use regression analysis, logistic regression, hypothesis testing, time series and differnt models to train my data.

Hypothesis Analysis on Marketing Campaign and Driving Miles

Analyze the SFO and LAX data sets and determine if either marketing campaign was successful in raising the average miles driven per Uber driver.

Logistic Regression on Customers Transaction Prediction

• Use logistic regression to predict when customers are going to transact
• Determine the causes for a transaction
• Evaluated the performance the model

Logistic Regression on Employee Turnover Prediction

• Use logistic regression to predict when people are going to leave a company
• Determine the causes for attrition
• Evaluated the performance the model

Predict Salary by Lasso & Ridge Regularization

• Use regularization to predict salaries for a sports player
• Explain the output of the regularized models

Regression Analysis on Housing Price

• Remove/manipulate/transform features from the data set, remain only useful data
• Graphically and numerically describe model performance and find the relation between them
• Apply regression analysis techniques and EDA principles to find out what features will influence the rental price

Telecom Customer Churn Prediction By Using Different Machine Learning algorithms

• Trialed a list of different Machine Learning algorithms, such as Logistic Regression(with Lasso & Ridge), Decision Tree, KNN Classifier, and Random Forest Classifier, and Linear Regression to predict potential customer churn and customer life time value.
• Provided the best model that has achieved the highest AUC value with lowest MSE(Mean Squared Error).
• Contructed the particial dependece plot to discover how the most 6 importance features related to the customer churn.

Time Series Analysis Using ARIMA on Electro Data Prediction

• Implemented ARIMA model, analyzed 2 data sets to predict the values for the next 8 time periods and the subsequent 7 years (with confidence intervals), and make 3 observations about the data (i.e., describe its composition and characteristics).

Forecasting on Video CTR

• Using Moving Average, Exponential smoothing, AR and ARIMA model to forecast video CTR (click through rate)
• Select a performance measure for the model and pick the best performing model with lowest MSE.

Multiple Regression Analysis on Civilization VI Game Players' Active Days

• Determine the causes of active day
• Use multiple regression model to predict players' active day