v0.5.0

0.5.0 (2023-09-28)

Features

• Add DataFrame.kurtosis / DF.kurt method (c1900c2)
• Add DataFrame.rolling and DataFrame.expanding methods (c1900c2)
• Add items, apply methods to DataFrame. (#43) (3adc1b3)
• Add axis param to simple df aggregations (#52) (9cf9972)
• Add index dtype, astype, drop, fillna, aggregate attributes. (#38) (1a254a4)
• Add ml.preprocessing.LabelEncoder (#50) (2510461)
• Add ml.preprocessing.MaxAbsScaler (#56) (14b262b)
• Add ml.preprocessing.MinMaxScaler (#64) (392113b)
• Add more index methods (#54) (a6e32aa)
• Support calculate_p_values parameter in bigframes.ml.linear_model.LinearRegression (c1900c2)
• Support class_weights="balanced" in LogisticRegression model (c1900c2)
• Support df[column_name] = df_only_one_column (c1900c2)
• Support early_stop parameter in bigframes.ml.linear_model.LinearRegression (c1900c2)
• Support enable_global_explain parameter in bigframes.ml.linear_model.LinearRegression (c1900c2)
• Support l2_reg parameter in bigframes.ml.linear_model.LinearRegression (c1900c2)
• Support learn_rate_strategy parameter in bigframes.ml.linear_model.LinearRegression (c1900c2)
• Support ls_init_learn_rate parameter in bigframes.ml.linear_model.LinearRegression (c1900c2)
• Support max_iterations parameter in bigframes.ml.linear_model.LinearRegression (c1900c2)
• Support min_rel_progress parameter in bigframes.ml.linear_model.LinearRegression (c1900c2)
• Support optimize_strategy parameter in bigframes.ml.linear_model.LinearRegression (c1900c2)
• Support casting string to integer or float (#59) (3502f83)

Bug Fixes

• Fix header skipping logic in read_csv (#49) (d56258c)
• Generate unique ids on join to avoid id collisions (#65) (7ab65e8)
• LabelEncoder params consistent with Sklearn (#60) (632caec)
• Loosen filter items tests to accomodate shifting pandas impl (#41) (edabdbb)

Performance Improvements

• Add ability to cache dataframe and series to session table (#51) (416d7cb)
• Inline small Series and DataFrames in query text (#45) (5e199ec)
• Reimplement unpivot to use cross join rather than union (#47) (f9a93ce)
• Simplify join order to use multiple order keys instead of string. (#36) (5056da6)

Documentation

• Link to Remote Functions code samples from README and API reference (c1900c2)

v0.4.0

0.4.0 (2023-09-16)

Features

• Add axis parameter to droplevel and reorder_levels (7c6b0dd)
• Add bfill and ffill to DataFrame and Series (7c6b0dd)
• Add DataFrame.combine and DataFrame.combine_first (#27) (7c6b0dd)
• Add DataFrame.nlargest, nsmallest (7c6b0dd)
• Add DataFrame.pct_change and Series.pct_change (7c6b0dd)
• Add DataFrame.skew and GroupBy.skew (7c6b0dd)
• Add DataFrame.to_dict, to_excel, to_latex, to_records, to_string, to_markdown, to_pickle, to_orc (7c6b0dd)
• Add diff method to DataFrame and GroupBy (7c6b0dd)
• Add filter and reindex to Series and DataFrame (7c6b0dd)
• Add reindex_like to DataFrame and Series (7c6b0dd)
• Add swaplevel to DataFrame and Series (7c6b0dd)
• Add partial support for Sereies.replace (7c6b0dd)
• Support DataFrame.loc[bool_series, column] = scalar (7c6b0dd)
• Support a persistent name in remote_function (7c6b0dd)

Bug Fixes

• remote_function uses same credentials as other APIs (7c6b0dd)
• Add type hints to models (7c6b0dd)
• Raise error when ARIMAPlus is used with Pipeline (7c6b0dd)
• Remove transforms parameter in model.fit (breaking change) (7c6b0dd)
• Support column joins with "None indexer" (7c6b0dd)
• Use for literals Int64Dtype in cut (7c6b0dd)
• Use lowercase strings for parameter literals in bigframes.ml (breaking change) (7c6b0dd)

Performance Improvements

• bigframes-api label to I/O query jobs (7c6b0dd)

Documentation

• Document possible parameter values for PaLM2TextGenerator (7c6b0dd)
• Document region logic in README (7c6b0dd)
• Fix OneHotEncoder sample (7c6b0dd)

v0.3.2

0.3.2 (2023-09-06)

Bug Fixes

• Make release.sh script for PyPI upload executable (#20) (9951610)

v0.3.1

0.3.1 (2023-09-05)

Bug Fixes

• release: Use correct directory name for release build config (#17) (3dd25b3)

v0.3.0

0.3.0 (2023-09-02)

Features

• Add bigframes.get_global_session() and bigframes.reset_session() aliases (a32b747)
• Add bigframes.pandas.read_pickle function (a32b747)
• Add components_, explained_variance_, and explained_variance_ratio_ properties to bigframes.ml.decomposition.PCA (89b9503)
• Add fit_transform to bigquery.ml transformers (a32b747)
• Add Series.dropna and DataFrame.fillna (8fab755)
• Add Series.str methods isalpha, isdigit, isdecimal, isalnum, isspace, islower, isupper, zfill, center (a32b747)
• Support bigframes.pandas.merge() (8fab755)
• Support DataFrame.isin with list and dict inputs (8fab755)
• Support DataFrame.pivot (a32b747)
• Support DataFrame.stack (89b9503)
• Support DataFrame-DataFrame binary operations (8fab755)
• Support df[my_column] = [a python list] (89b9503)
• Support Index.is_monotonic (8fab755)
• Support np.arcsin, np.arccos, np.arctan, np.sinh, np.cosh, np.tanh, np.arcsinh, np.arccosh, np.arctanh, np.exp with Series argument (89b9503)
• Support np.sin, np.cos, np.tan, np.log, np.log10, np.sqrt, np.abs with Series argument (89b9503)
• Support pow() and power operator in DataFrame and Series (8fab755)
• Support read_json with engine=bigquery for newline-delimited JSON files (89b9503)
• Support Series.corr (89b9503)
• Support Series.map (8fab755)
• Support for np.add, np.subtract, np.multiply, np.divide, np.power (8fab755)
• Support MultiIndex for DataFrame columns (a32b747)
• Use pandas.Index for column labels (a32b747)
• Use default session and connection in ml.llm and ml.imported (8fab755)

Bug Fixes

• Add error message to set_index (a32b747)
• Align column names with pandas in DataFrame.agg results (89b9503)
• Allow (but still not recommended) ORDER BY in read_gbq input when an index_col is defined (89b9503)
• Check for IAM role on the BigQuery connection when initializing a remote_function (89b9503)
• Check that types are specified in read_gbq_function (a32b747)
• Don't use query cache for Session construction (a32b747)
• Include survey link in abstract NotImplementedError exception messages (89b9503)
• Label temp table creation jobs with source=bigquery-dataframes-temp label (89b9503)
• Make X_train argument names consistent across methods (8fab755)
• Raise AttributeError for unimplemented pandas methods (89b9503)
• Raise exception for invalid function in read_gbq_function (a32b747)
• Support spaces in column names in DataFrame initializater (89b9503)

Performance Improvements

• Add local cache for __repr_*__ methods (a32b747)
• Lazily instantiate client library objects (89b9503)
• Use row_number() filter for head / tail (8fab755)

Documentation

• Add ML section under Overview (a32b747)
• Add release status to table of contents (a32b747)
• Add samples and best practices to read_gbq docs (a32b747)
• Correct the return types of Dataframe and Series (a32b747)
• Create subfolders for notebooks (a32b747)
• Fix link to GitHub (89b9503)
• Highlight bigframes is open-source (a32b747)
• Sample ML Drug Name Generation notebook (a32b747)
• Set options.bigquery.project in sample code (89b9503)
• Transform remote function user guide into sample code (a32b747)
• Update remote function notebook with read_gbq_function usage (8fab755)

Version 0.2.0

0.2.0 (2023-08-17)

Features

• Add KMeans.cluster_centers_.
• Allow column labels to be any type handled by bq df, column labels can be integers now.
• Add dataframegroupby.agg().
• Add Series Property is_monotonic_increasing and is_monotonic_decreasing.
• Add match, fullmatch, get, pad str methods.
• Add series isin function.

Bug Fixes

• Update ML package to use sessions for queries.
• Optimize read_gbq with index_col set to cluster by index_col.
• Raise ValueError if the location mismatched.
• read_gbq no longer uses 'time travel' with query inputs.

Documentation

• Add docstring to _uniform_sampling to avoid user using it.

Version 0.1.1

Documentation

• Correct link to code repository in setup.py and use correct terminology for
  console.cloud.google.com links.

Version 0.1.0

0.1.0 (2023-08-11)

Features

• Add bigframes.pandas package with an API compatible with
  pandas. Supported data sources include:
  BigQuery SQL queries, BigQuery tables, CSV (local and GCS), Parquet (local
  and Cloud Storage), and more.
• Add bigframes.ml package with an API inspired by
  scikit-learn. Train machine learning
  models and run batch predicition, powered by BigQuery
  ML.