Skip to content

TTF DataScience

Overview

The mission of the TTF Data Science Team is to help develop solutions for data-related issues to support the research efforts of the NCCR Evolving Language.

Our services include (but are not limited to):

  • Database building and data processing
  • Machine learning solutions
  • Statistical analysis and modelling

See Services below for more information.

How to request for support

To request for support from the TTF DataScience team, please write an email to datascience_request_AT_evolvinglanguage.ch and always include:

  • Your Task/WP/SIG and affiliation
  • Your role in the NCCR (e.g., PhD, PostDoc, …)
  • A brief description of your issue

Important: If you are a PhD student, please make sure that you communicate with your supervisors before writing to us and specify to which part (chapter/paper) of your dissertation this request is related.

Please also note that you must follow the NCCR authorship guidelines if any of our team members is involved in your publications.

Team

Principle investigators

Richard Hahnloser, Sabine Stoll, Reinhard Furrer

Staff members

Services

Databases

The WP Databases supports researchers in the use and the management of their data throughout the data life cycle. We help facilitate the data management of NCCR to adhere to the SNF data management requirement. Therefore, we monitor and maintain the dataset index of the NCCR, as well as major databases of the NCCR such as ACQDIV and VoCallBase.

In addition, our WP is also at your disposal for specific tasks. Upon request, we provide services including:

  • Consulting on data processing, database operations and design, and general data analysis.
  • Technical support for general data-related issues, such as building data pipelines for more efficient data processing (e.g., data cleaning, formatting and structuring for data from multiple sources) and designing applications/services for data visualization.
  • Data management solutions such as data security issues and archiving instructions

Machine Learning

The WP Machine Learning provides support to researchers on problems and tasks related to machine learning or deep learning. Upon request, our services include (but are not limited to) the following:

  • Consulting on general machine learning issues is available. For example, we can suggest possible machine learning or deep learning frameworks for your data analysis, such as clustering, classification, regression or data pre-processing to make it ready for down-streaming tasks.
  • Technical support for machine learning models using modern frameworks such as PyTorch or Tensorflow. We can either help you diagnose your model or existing code, or we can build a tool using cutting-edge techniques based on your requirements.
  • We can also be involved in research topics that are of broad interest to multiple research groups, such as syllable segmentation and speech processing, and play a major role in developing methods and providing solutions.

Statistics

The WP Statistics supports researchers in all types of statistical tasks, ranging from study design to the development of new methods. Our services can be grouped into three levels, listed below in order from least to most time-intensive.

  • Chat-istician: for when you need to talk through the statistical aspects of study design or analysis (e.g., your analysis plan for a grant or pre-registration, model comparison, power analysis, inference). This might involve directing you to relevant resources (e.g., R packages, tutorials, methods papers), or helping you respond to peer reviewers.
  • Technical Support: for when you need someone to diagnose why your model results seem “off”, or why your code is broken. This level is more hands-on, and can take more time. It may involve code review and working through mathematical or computational issues together.
  • Collaboration: for when you need a statistician to join your paper or research project. At this level, the consulting statistican plays a major role in the design and/or implementation of data analysis. Depending on the scope of the work, level three may entail co-authorship.
We suggest that researchers use these levels as heuristics for communicating the scope of the support that they need. In practice, the boundaries between these levels can blur. A request that starts as a simple consulting chat may develop into a long-term collaboration, and we are happy to guide you through each stage.