data-gym: Demo notebooks & datasets

Listing of datasets

General

Data I/O

• Extracting data: background color of cells in Excel files
• Reading in data from Google spreadsheet
  • Example 2: Subject999
• Converting from PDF using batch mode

Plotting

• Creating subplots
• Correlogram with numbers
• Example of plotting in 3D

Prelim. analyses

• Discretize variables using numpy.matlib.repmat
• Data visualization via dimensionality-reduction
• Explore some ONC datasets
• Explore the YaleED dataset
• Explore clinical codes; calculate some clinical indices

Preprocessing

• biosppy

Training models

• Logistric regresssion and LSTM using PIMA
• LR and MLP using PIMA
• Time-series classification (WISDM dataset)
  • MLP
  • LSTM, GRU
  • 1D-CNN

Misc.

• Improving reproducibility with seeds

Common code snippets

• Jupyter-oriented

Helpful links

• Start R notebook
• Getting started with R by me

Resources

Visual analytics

• "Making Data Visual" by Danyel Fisher, Miriah Meyer
  • How to choose?
• Visualizing missing data quickly
• Visualization: bad examples
• Violinplot via Seaborn
• Cartograms (explained in first ten seconds)

Statistics & cross-validation

• "Statistical Approaches to the Model Comparison Task in Learning Analytics," Gardner & Brooks, LAK2017
• Slides "Statistical testing for classification..." by B Evans
• Notes: "[...] ordinal regression is known to perform better than softmax on ordinal data (Cheng et al., 2008)"

Meta-data

• BioLinCC
• DataOne
• DataOne search interface