Seeking a position of Data Scientist or Research Scientist
I hold a PhD in computational neuroscience and am currently a Data Incubator fellow. I developed predictive behavioral models in my PhD work. At the Data Incubator, I have built a model that effectively predicts loan outcome. I have also built my own data processing pipeline and have extensive experience with MatLab, python, and SQL (See Resume). I am looking for a full-time Data Scientist position.
On average, 18% of loans in Lending Club go to default. I developed an ensemble model (including gradient boosting decision tree, naive bayse and logistic regression) to help people spot those faulty loans, reducing investment risk by 50%.
I initially developed this network as a graduate school class project. It is a multilayer perceptron used for recognizing handwritten digits. Training is fulfilled by using a back-propagation algorithm. I used a popular dataset (‘mnist_all.dat’) comprising of training and testing samples of the different digits is provided in the Midterm folder. Each sample is a 28x28 gray scale 8-bit image.
I built a random forest model to predict the monthly unadjusted Canadian retail sales number. The prediction explains 98.66% of the variance in the actual, historical data. In almost every case (probability of 0.95), the error in the prediction (the absolute difference between the actual data and the prediction) is less than ~5% of the average sales number.
The goal of this analysis is to assess whether one process drives the correlation between two other signals. Ideally, one should extract just the shared (correlated) component of the signals, and then use causality analysis (e.g., Granger causality analysis or transfer entropy) to determine whether the process of interest drives that correlated component. Methods for extracting the correlated component require a large number of correlated signals as input to operate properly. With small amounts of correlated signals, the extracted component will necessarily include a substantial amount of non-correlated signal, regardless of the precise method used. Regression based dependency analysis circumvents the need for extracting only the correlated component.
Results are published in 'Li J, Bentley W, Snyder L. Neural Correlates of Functional Connectivity. Organization for Human Brain Mapping, 2015.'
A python project of the model described in Honey et.al 2007 PNAS, with expansion to capture both ultra-slow (minutes) and ultra-fast (millisecond) neural dynamics.
Li JM, Bentley WJ, Snyder AZ, Raichle M, Snyder LH (2015) Functional connectivity arises from a slow rhythmic mechanism, Proc Natl Acad Sci U S A. In press.
Bentley WJ*, Li JM*, Snyder AZ, Raichle M, Snyder LH (2014) Oxygen level and LFP in task positive and task negative areas: Bridging BOLD fMRI and electrophysiology. Cerebral Cortex: In press. (* Equal contribution)
Kubanek J, Li JM, Snyder LH (2015) Motor role of parietal cortex in a monkey model of hemi-spatial neglect, Proc Natl Acad Sci U S A. In press.
Li J, Bentley W, Snyder L. Functional Connectivity in Unit Activity. Organization for Human Brain Mapping, 2015.
Li J, Bentley W, Snyder L. Neural Correlates of Functional Connectivity. Organization for Human Brain Mapping, 2015.
Li J, Bentley W, Snyder L. Neural activity underlying functional connectivity MRI. Society for Neuroscience, 481.01. 2014
Bentley WJ, Li J, Snyder A, Raichle M, Snyder L. Functional connectivity arises from a slow rhythmic mechanism. Society for Neuroscience, 481.03. 2014
Bently W, Li J, Snyder A, Raichle M, Snyder L. Bridging BOLD fMRI and Electrophysiology: Oxygen polarography in awake macaques. Organization for Human Brain Mapping, 2013.
Bently W, Li J, Snyder A, Raichle M, Snyder L. Oxygen polarography and electrophysiology in the default-mode and dorsal-attention networks during rest and stimulation: Bridging BOLD fMRI and electrophysiology. Society for Neuroscience, 586.05. 2013
| Neural activity underlying functional connectivity MRI | 2014 |
| Society for Neuroscience Conference | |
| Washington, DC | |
| Neural mechanisms of functional connectivity MR imaging | 2014 |
| Anatomy and Neurobiology Department Trainee Seminar | |
| St Louis, MO | |
| Oxygen level and local field potential in a default mode area: | 2013 |
| Bridging BOLD fMRI and electrophysiology | |
| Society for Neuroscience Conference | |
| San Diego, CA | |
| Local Oxygen Fluctuations in the Brain | 2012 |
| Cognitive, Computational and Systems Neuroscience Symposium | |
| St Louis, MO |
| Ph.D. | Neuroscience (Computational) | Washington University in St. Louis | 2015 |
| B.S. | Biotechnology | Peking University, China | 2010 |
Data Science Fellow, the Data Incubator |
2015-present |
Consultant, the BALSA group |
2014-2015 |
Patent analysis involving a range of fields, including next generation DNA sequencing, bio-sensing nanoparticles, and cancer therapy
Graduate Researcher/Postdoctoral Fellow |
2010-present |
Teaching Assistant |
2011 |
NeuroDay Presenter, Saint Louis Science Center |
2011 |
Director of Organization Department, Peking University Chinese Literature Club |
2008 |
Summer Olympic and Special Olympic Volunteer |
2008 |
C, Python, MATLAB, SQL, Scala, Hadoop Mapreduce, Spark
Machine learning, Scikit Learn, decision tree, naive bayesian, artificial neural networks, natural language processing, web scraping, statistical analysis, probabilistic theory, information theory
Unix/Linux (Shell scripting)
| Finalist for O'Leary Prize | 2015 |
| (Award recognizing outstanding dissertation research) | |
| Acceptance into Cognitive, Computational, and Systems Neuroscience pathway | 2014 |
| (National Institutes of Health funded training program) | |
| Ellen Eoyang Scholarship | 2008 |
| (Award recognizing outstanding academic performance) | |
| Outstanding Olympics Volunteer Award | 2008 |
| 3rd Prize of China Biology Olympiad (Provincial Level) | 2005 |
| 2rd Prize of China Mathematical Olympiad (Provincial Level) | 2000 |
Jingfeng Li
jingfengmli@gmail.com
jingfengli@go.wustl.edu
314-616-2743