Sam

Changes
- `sam.feature_engineering.decompose_datetime` 'components' argument now support 'secondofday'

Changes
- `sam.visualization.quantile_plot` 'score' argument changed to 'title' to enhance generalizability

New functions
- New function: `sam.visualization.quantile_plot` function creates an (interactive) plot of SamQuantileMLP output

Changes
- `sam.feature_engineering.decompose_datetime` now has an new argument 'onehots' that converts time variables to one-hot-encoded
- `sam.feature_engineering.BuildRollingFeatures`: now as an argument 'add_lookback_to_colname'
- `sam.models.SamQuantileMLP`: now has argument 'time_onehots', default time variables adjusted accordingly
- `sam.models.SamQuantileMLP`: now has argument 'y_scaler'

Bugfixes
- `sam.models.SamQuantileMLP`: setting use_y_as_feature to True would give error if predict ahead was 0.

New functions
- New function: `sam.models.create_keras_autoencoder_mlp` function that returns keras MLP for unsupervised anomaly detection
- New function: `sam.models.create_keras_autoencoder_rnn` function that returns keras RNN for unsupervised anomaly detection
- Change `sam.models.create_keras_quantile_mlp`: supports momentum of 1.0 for no batch
normalization. Value of None is still supported.
- Change`sam.models.create_keras_quantile.rnn`: supports lower case layer types 'lstm' and 'gru'

A lot changed in version 2.0.0. Only changes compared to 1.0.3 are listed here.
For more details about any function, check the documentation.

New functions

- `sam.preprocessing.RecurrentReshaper` transformer to transform 2d to 3d for Recurrent Neural networks
- `sam.preprocessing.scale_train_test` function that scales train and test set and returns fitted scalers
- `sam.validation.RemoveFlatlines` transformer that finds and removes flatlines from data
- `sam.validation.RemoveExtremeValues` transformer that finds and removes extreme values
- `sam.validation.create_validation_pipe` function that creates sklearn pipeline for data validation
- `sam.preprocessing.make_differenced_target` and `sam.preprocessing.inverse_differenced_target` allow for differencing a timeseries
- `sam.models.SamQuantileMLP` standard model for fitting wide-format timeseries data with an MLP
- `sam.models.create_keras_quantile_rnn` function that returns a keras RNN model that can predict means and quantiles
- Functions for benchmarking a model on some standard data (in sam format): `sam.models.preprocess_data_for_benchmarking`,
`sam.models.benchmark_model`, `sam.models.plot_score_dicts`, `sam.models.benchmark_wrapper`
- `sam.feature_engineering.AutomaticRollingEngineering` transformer that calculates rolling features in a smart way

New features

- `sam.data_sources.read_knmi` has an option to use a nearby weather station if the closest weather station contains nans
- `sam.exploration.lag_correlation` now accepts a list as the `lag` parameter
- `sam.visualization.plot_lag_correlation` looks better now
- `sam.recode_cyclical_features` now explicitly requires maximums and provides them for time features
- Added example for SamQuantileMLP at `http://10.2.0.20/sam/examples.html#samquantilemlp-demo`

Bugfixes

- `sam.preprocessing.sam_format_to_wide` didn't work on pandas 0.23 and older
- `sam.exploration.lag_correlation` did not correctly use the correlation method parameter
- `sam.metrics.keras_tilted_loss` caused the entire package to crash if tensorflow wasn't installed
- `sam.visualization.plot_incident_heatmap` did not correctly set the y-axis
- `sam.feature_engineering.BuildRollingFeatures` threw a deprecationwarning on newer versions of pandas
- General fixes to typos and syntax in the documentation

Added new functions: `keras_joint_mse_tilted_loss`, `create_keras_quantile_mlp`