lazyrdatasets 0.2.0

Convenience interface of Reatasets for lazy data scientists.

lazy-rdatasets

Convenience interface of Rdatasets for lazy data scientists.

Key features

• Search and filter datasets by text, data types, no. samples, and no. features/parameters
• Randomly select a dataset
• Easily do basic visualizations without writing many lines

Install

Use pip:

pip install lazy-rdatasets

Usage

List all available datasets

from lazyrdatasets import LazyRdatasets

rd = LazyRdatasets()
rd

   Package  Item           Title                              Rows  Cols  ...
0  AER      Affairs        Fair's Extramarital Affairs Data    601     9  ...
1  AER      ArgentinaCPI   Consumer Price Index in Argentina    80     2  ...
2  AER      BankWages      Bank Wages                          474     4  ...
3  AER      BenderlyZwick  Benderly and Zwick Data: Infla...    31     5  ...
4  AER      BondYield      Bond Yield Data                      60     2  ...
...

Finding datasets

1. Exact match on the package and item names

This is almost the same as statsmodels.datasets.get_rdatasets.

from lazyrdatasets import LazyRdatasets

rd = LazyRdatasets.find(package="palmerpenguins", item="penguins", exact=True)

2. Search a string in the title of datasets

Find datasets that have "penguin" in their title:

from lazyrdatasets import LazyRdatasets

rd = LazyRdatasets.find(title="penguin")

3. Filtering datasets that contains particular types of variables

Find datasets that have categorical variables:

from lazyrdatasets import LazyRdatasets

rd = LazyRdatasets.find(categorical=True)

Find datasets that have only one numeric variables with more than 99 samples:

from lazyrdatasets import LazyRdatasets

rd = LazyRdatasets.find(numeric=True, nmin=100, pmax=1)

Getting a single dataset from Rdatasets

Below, rd is the output of the following code block:

from lazyrdatasets import LazyRdatasets

rd = LazyRdatasets.find(title="penguin")

1. Show the list of datasets that matched the conditions

rd  # On jupyter, this will show the same result as below.
rd.catalog  # Show the repr of DataFrame

      Package          Item                     Title   ...
567   bayesrules       penguins_bayes           Penguins Data
1694  heplots          peng                     Size measurements ...
2116  modeldata        penguins                 Palmer Station penguin data
2559  palmerpenguins   penguins                 Size measurements for  ...
2560  palmerpenguins   penguins_raw (penguins)  Penguin size, clutch, ...

2. Pick the first dataset in the list

rd.first

#️⃣ Index  : 567
📦 Package: bayesrules
📄 Item   : penguins_bayes
📚 Title  : Penguins Data
📐 Shape  : (344, 10)
  ⚖️ Binary   : 2
  🔤 Character: 0
  🧮 Factor   : 4
  🔘 Logical  : 0
  🔢 Numeric  : 5
🔗 CSV: https://vincentarelbundock.github.io/Rdatasets/csv/bayesrules/penguins_bayes.csv
🔗 Doc: https://vincentarelbundock.github.io/Rdatasets/doc/bayesrules/penguins_bayes.html

3. Acceess by Index and position in the catalog

rd[2559]  # Get the Dataset with its index 2559 (pandas.DataFrame.loc), or ...
rd.at(3)  # Get the Dataset at position 3 in the catalog (pandas.DataFrame.iloc)

#️⃣ Index  : 2559
📦 Package: palmerpenguins
📄 Item   : penguins
📚 Title  : Size measurements for adult foraging penguins near Palmer Station, Antarctica
📐 Shape  : (344, 9)
  ⚖️ Binary   : 1
  🔤 Character: 0
  🧮 Factor   : 3
  🔘 Logical  : 0
  🔢 Numeric  : 5
🔗 CSV: https://vincentarelbundock.github.io/Rdatasets/csv/palmerpenguins/penguins.csv
🔗 Doc: https://vincentarelbundock.github.io/Rdatasets/doc/palmerpenguins/penguins.html

Getting the dataframe

ds = rd[2559]
ds.data  # -> pandas.DataFrame

   rownames species     island  bill_length_mm  bill_depth_mm ...
0         1  Adelie  Torgersen            39.1           18.7
1         2  Adelie  Torgersen            39.5           17.4
2         3  Adelie  Torgersen            40.3           18.0
3         4  Adelie  Torgersen             NaN            NaN
4         5  Adelie  Torgersen            36.7           19.3
...

Quicklook

Selected dataset can be easily quicklooked by

ds.quicklook()

Missing Values	Categorical Variables	Numeric Variables(p≧1)	Numerical Variables (p≧3)
Heatmap	Bar plots	Scatter matrix (p≧2) / histogram (p=1)	PCA projection (p≧3)

This is of course not the best visualization, but might be helpful to get an overview of the dataset.