Table 3 Feature engineered predictors and their description
Calendar feature name | Description | Calendar feature type |
|---|---|---|
date_index.num | Time is converted into a numerical value in seconds from a fixed base date set at 2014–01-01 00:00:00 = 0, where 2014–01–02 corresponds to 86,400 s, 2014–01–03 = 172,800, and so on. The variable represents the number of seconds elapsed from 2014–01-01 to 2016–01–31 | Numeric |
date_half | The variable indicates whether the date falls in the first or second half of the year (e.g., 2014–01-01 = 1, 2014–07-01 = 2) | Categorical [1 or 2] |
date_quarter | It represents the quarterly component of the index. The year is divided into four quarters, each including three consecutive months. The variable indicates to which quarter of the year a specific date belongs (e.g., January 15 = 1; April 28 = 2), enabling data analysis based on quarterly patterns or trends | Categorical [1 to 4] |
date_mday | The variable indicates the day of the month associated with a particular date (e.g., January 15 = 15) | Categorical [1 to 31] |
date_qday | The variable represents the day of the quarter, ranging from 1 to 92 for a given date, with each quarter including from 90 to 92 days (e.g., June 30 = 91, as it is the 91st day of the second quarter; September 30 = 92, as it is the 92nd day of the third quarter) | Categorical [1 to 92] |
date_yday | The variable represents the day of the year, ranging from 1 to 365, for a given date (e.g., March 31 is the 90th day of the year), enabling analysis and grouping of data based on annual patterns or trends | Categorical [1 to 365] |
date_mweek | The variable represents the week of the month, ranging from 1 to 5, for a given date (e.g., January 7 = 1; January 15 = 3) | Categorical [1 to 5] |
date_week | The variable represents the week number of the year (considering the first week starts on the first Sunday). Thus, in a year where January 1st falls on a Tuesday, this week is designated as week 53 of the previous year. Week 1, in turn, begins on January 6th | Categorical [1 to 53] |
date_week2 | The variable is a binary indicator representing the biweekly frequency module. The term “module” refers to the number that represents two possible states in each two-week cycle, taking values of 1 or 0 (e.g., January 7 = 1; January 14 = 0; January 15 = 1) | Binary [1 or 0] |
date_week3 | The variable represents the three-week frequency module. The variable can take on values of 1, 2, or 0 (e.g., January 7 = 1; January 14 = 2; January 15 = 0; and January 22 = 1) | Categorical [0 to 2] |
date_week4 | The variable represents the quadriweekly frequency module, with values ranging from 0 to 3 (e.g., January 7 = 1; January 14 = 2; January 15 = 3; January 22 = 0; and January 29 = 1) | Categorical [0 to 3] |
date_mday7 | The variable is used to order each weekday occurrence within a month, starting from 1 (e.g., the first Saturday of the month will have mday7 = 1, the second mday7 = 2, and so on; the same applies to other weekdays) | Categorical [1 to 5] |