Astronomy of the day is more effective than seasonal decomposition in modeling and predicting the rates of Amazon review misspellings

Astronomy

Authors

Vol. 7 No. 02 (2019)
Astronomy and Astrophysics
February 8, 2019

Downloads

This study shows that the misspelling rates in Amazon reviews can be effectively modeled as having cyclical components by using seasonal decomposition, and this effectiveness is almost identical to that gained by using a neural network that uses basic astronomical placements of the day. While the modeling of past Amazon misspelling rates was slightly more effective using a neural net and astronomy of day as compared to seasonal decomposition, future values were only effectively predicted by the neural net that used astronomy of day.