v1.0.17

This release removes dependency on the now deprecated 'Assertive' package. This was described in issue #27. Thanks to everyone who pointed this out.

v1.0.16

This release fixes a bug in monthly change point models (introduced in v1.0.15) that resulted in apparent overfitting of monthly models.

Updates in this release:

The function 'model_with_cp` was updated to:

• Remove interval_start and interval_end from the training data for monthly models
• Change the columns that are kept inside the function when day_normalized is TRUE

This was described in Issue #28. Thanks to Nastaran Alishahi for pointing this out!

v1.0.15

Updates in this release:

The aggregate and create_dataframe functions have been re-written to accomplish these main tasks:

• Fix time shifting issues
• Remove dependence on the xts package
• Add support for additional independent variables to the aggregate function
• Increase performance
• Simplify the code, increase readability, and improve documentation
• Unify the procedures used for aggregation
• Closes issue #14

Part of the procedure unification process involved stripping the primary processing code out of the create_dataframe function and instead calling to the aggregate function to serve as the processing engine. create_dataframe now acts as a user friendly front end function that will check and error trap inputs. aggregate can also be called directly by nmecr users, but its purpose is more intended to be an procedure that can be used within other functions and analysis scripts where the developer is certain that arguments are formatted correctly.

Additional changes included:

• Remove while loop in five parameter change point model that was causing model fit task to get stuck in an infinite loop
• Fix typos in the vignettes
• Update author list

v1.0.14

Updates in this release:

• Update calculate_coverage to explicitly use numeric version of bins object. Fixes error caused by newer version of dplyr.

v1.0.13

Updates in this release:

• Removing dependency on the timetk package.

v1.0.12

Updates in this release:

• Added monthly model support for weather coverage calculation
• Fixed issue #15 due an an assignment operator bug
• Fixed issue #16 by adding an argument to calculate_model_predictions to allow for negative values to be predicted using TOWT

v1.0.11

Update in this release:

• Resolved an infinite loop issue that was caused when picking starting temperatures too close to the boundary temperatures for the 5P model. This was related to Issue #11.

v1.0.10

Updates in this release:

• corrected naming conventions for normalized savings summary, calculation of model parameters, and normalized savings datapoint count
• updated vignettes to match parameter correction and normalized savings correction

v1.0.9

Updates in this release:

• Added in unit tests for basic dataframe creation checks and vignette checks
• corrected 3PC and 3PH model forms to restrict the algorithm and force 3P models when requested
• Added in warning messages for when 3PC and 3PH cannot be computed (due to lack of appropriate changepoint)
• Corrected day normalization calculation for monthly dataframes
• Documentation updates

v1.0.8

Updates in this release:

• All algorithms can have additional variables for modeling as well as prediction
• Mean model
• Allowing inputs of less than 15-min interval data
• Minor correction in weather coverage
• Use of periodicity of timeseries directly instead of scale to determine data intervals for further calculations