Date - 09/16/2020
Release Tag - pd.0.0.2
Release Status - Current head of branch master
Update to use datacommons Python API 1.4.2, which adds batching to the get_stat_all function used by build_time_series_dataframe and build_multivariate_dataframe.
Date - 08/25/2020
Release Tag - pd.0.0.1
Release Status - Current head of branch master
Added pandas wrapper functions.
build_time_series
constructs a pd.Series for a given StatisticalVariable and Place, where the time series are indexed by date.build_time_series_dataframe
constructs a pd.DataFrame for a given StatisticalVariable and a set of Places. The DataFrame will have Places as the index and dates as the columns.build_multivariate_dataframe
constructs a pd.DataFrame for a set of StatisticalVariables and a set of Places. The DataFrame will have Places as index and StatisticalVariables as the columns. The values are the most recent values for the chosen StatVarObservation options.
For multi-place functions, when a StatisticalVariable has multiple StatVarObservation options, Data Commons chooses a set of StatVarObservation options that covers the most places. This ensures that the data fetched for a StatisticalVariable is comparable across places. When there is a tie, we select the StatVarObservation options set with the latest date data is available for any place.