The Trans-Atlantic Database includes 34,948 slave trade voyages from the mid-16th to mid-19th centuries. The database contains 256 variables including the number of enslaved people embarked and disembarked and the major ports at which they embarked and disembarked. We explored the correlation between the death rate and the number of enslaved people embarked, the variation of the death rate over time, and the distribution of the death rate using this database (view Analyze Mortality Patterns).
Although the Trans-Atlantic Database provides a large amount of information, it lacks day-to-day geographic records and has many missing values. Therefore, the CLIWOC database which contains detailed geographic coordinates and/or weather events for ships (including but not limited to slave ships) from 1750 to 1850 is used as a supplement to the Trans-Atlantic Database to find and predict the Trans-Atlantic Database’s missing voyage paths.
In order to find and predict voyage paths in the Middle Passage, we merged the two databases mentioned above based on ship names and found 316 ships with the same names, of which we deduced 35 matching voyages. For voyages in both databases, we referred to ship logs and historical archives to find where and how enslaved people died (view Locate Trans-Atlantic Mortality). For voyages only in the Trans-Atlantic Database, we used differential equations and recurrent neural network to predict voyage paths (view Predict Voyage Paths). Figure 1 is the workflow we used to find and predict the slave trade voyage paths in the Middle Passage.