Release 0.6.0

Changes:

Interface:

• y must now be a matrix of either Tx1 or TxP.
• trace specifies whether a progress bar will be printed or not. Default to TRUE.
• loss_function lets you now specify "quantile", "expectile" or "percentage". All functions are generalized as in Gneitling 2009. The power can be scaled by loss_parameter. The latter defaults to 1, which leads to the well-known quantile, squared, and absolute percentage loss.
• gradient lets you specify whether the learning algorithms should consider actual loss or a linearized version using the gradient of the loss. Defaults to TRUE (gradient-based learning).
• forget_performance was added. It defines the share of the past performance that will be ignored when selecting the best parameter combination.
• Renamed forget parameter to forget_regret to underline its reference to the regret.
• New init_weights parameter. Has to be either a Kx1 or KxP matrix specifying the experts' starting weights.
• Add lead_time parameter. offset for expert forecasts. Defaults to 0 which means that experts predict t+1 at t. Setting this to h means experts predictions refer to t+1+h at time t. The weight updates delay accordingly.

Internal:

• If more expert forecasts than observations are provided, the excessive expert forecasts are used for prediction using the most recent weights.
• tau is now optional. It defaults to 1:P/(P+1). A scalar given to tau will be repeated P times. The latter is useful in multivariate settings.
• The pinball_loss and loss_pred functions were replaced by a more flexible function called loss.
• The weights object is changed from a T+1xKxP array to a T+1xPxK array to match other objects' dimensions. Now the following indexing scheme is consistent throughout the package: (Time, Probabilities, Experts, Parameter combination)
• Fixed Bug that caused single quantile calculations to fail.
• Various internal changes to improve readability and performance.

Refer to NEWS.md for changes in older versions.