Free Access

Table 3.

Summary of the advantages and disadvantage of different models considered in this work.

Property OLS LASSO RF Bayesian
Non-linear No No Yes Yes
Interpretable Yes Yes Yes (a) Yes
Suitable for all data sizes? Yes Yes Mid-sized to big data Small to mid-sized data
Interactions No No (b) Yes Yes
Prediction of trends No No No Yes
Poor signal to noise ratio No No Yes Yes

Notes.

(a)

Feature importances and partial dependence plots can help interpret random forests.

(b)

Including cross terms like can help model interactions but it still assumes each feature is linearly independent.

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.