Jkgqwsbblksnugwielwy

Chung-Fu(Alger), Yang

Data Scientist | Data Analyst | Statistician | Machine Learning Engineer

[email protected] 

+886-956-311-766 / Taipei, Taiwan

Working Experience

PEGATRON, Data Scientist/Researcher, Feb 2018 - Present

Optimized Operating Method for Chiller System and Automatically control in real time

  • Design objective function for energy efficiency saving problem
  • Design Algorithm with data driven and control chiller system automatically
  • Co-work with customer and domain expert in Shanghai to get insignts
  • Develop Notification System to send mail automatically to improve energy consumption and avoid chiller or relative devices broken
  • Build machine learning model such as LGBM/XGBoost/NN to predict temperture and power for each chillers and choose the best control strategy 
  • Saving about 500 thousand RMB per years 

Automatically Production Line Analysis System in real time:

  • Detect combination of abnormal material vendors with statistical testing
  • Create recommended lists which combinations of vendors are better/worse than others and improve productivity loss effectively
  • Weekly usage achieve 200 ~ 300 times & Improve Yield Rate 
Experimental design for manufactory to choose the better label tools :
  • Design Sampling method and Experiment to get user behavior data 
  • Be a contact to communicate with project lead in manufactory to make them understand the experiment and instruct operater to complete experiment
  • Nonparametric statistics and ANOVA in AB-test to recommend the most suitable label tools make senior manager can make decision
Operator Allocation System : 
  • Honor : Taiwan and Other Countrys patent with Deep Learning (pending)
  • Find the best working Station for each worker to reduce Fail rate
  • Data ETL & design the core algorithm by machine learning/ linear programming
  • Design simulation & AB test experiment to evaluate model performance
Deep learning training program for PEs and managers : 
  • As a mentor : Instruct participants to solve problems according to their project
  • As a host : Be a host in Pegatron AI training program closed day
Experiences 00 00@2x

DSP Company, Data Analyst Intern, Apr 2016 - Sep 2016

  • Co-work with domain experts and journalist in THE STORM MEDIA 
  • Analyze the open data of power in Taipei and explore user behavior
  • Design machine learning model to explain user behavior and insight
  • Online news report :『 Who is high-power user?
  • Project sharing talk activity
Lecturer of Data Analysis Lecture with R 
Experiences 00 01@2x

Honor & Award

Kaggle Competitions Expert, top1%, (2,142/152,285)

Mechanisms of Action (MoA) Prediction, top4%, 162/4,373

  • Well designed training process including feature design, PCA, Clustering
  • Design the customized variant NN/ ResNet/ TabNet network 
  • Apply transfer learning with NN/ ResNet/ TabNet in Multi-Label classification
  • Ensemble strategy with Lagrange Multiplier to Optimise Blending Weights
  • Well generalized model by training/testing with model ensembling and label smoothing 

RSNA STR Pulmonary Embolism Detection, top8%, 87/987

  • Design GRU to Classify Pulmonary Embolism cases in chest CT scans

Champion of 2016 Aisa Open Data Hackathon

Education

Master of Statistics, National Chengchi University, Sep 2014 - Jun 2016

Bachelor of Science in Mathematics, Soochow University, Sep 2010 - Jun 2014


Research/Model                         Skill


Deep Learning

  • CNN, MLP, Transfer Learning, Keras for tabular data 

Machine Learning

  • XGBoost, Neural Network, SVM, KNN, K-Means, LGBM, Tree-Based Model, PCA
Statistical Methods 
  • Experimental Design (ANOVA) 
  • Applied Linear Regression Machine Learning 
Problem Formulation
  • transfer real-world problem to mathematical model 
  • design customized models and training process 


Programming

  • R : tidyverse, dplyr, rmarkdown, ggplot, shiny
  • Python : scikit-learn, pandas, numpy, Keras
  • Database : mysql, nosql(mongo)
Others
  • linux (ubuntu)
  • docker
  • git
Language
  • TOEIC : 820