{"id":2701,"date":"2022-11-19T01:33:38","date_gmt":"2022-11-19T01:33:38","guid":{"rendered":"http:\/\/optimumsportsperformance.com\/blog\/?p=2701"},"modified":"2022-11-19T15:06:13","modified_gmt":"2022-11-19T15:06:13","slug":"tidymodels-train-test-set-without-initial_split","status":"publish","type":"post","link":"https:\/\/optimumsportsperformance.com\/blog\/tidymodels-train-test-set-without-initial_split\/","title":{"rendered":"tidymodels train\/test set without initial_split"},"content":{"rendered":"

Introduction<\/strong><\/p>\n

In {tidymodels<\/strong>} it is often recommended to split the data using the initial_split()<\/strong> function. This is useful when you are interested in a random sample from the data. As such, the initial_split()<\/strong> function produces a list of information that is used downstream in the model fitting and model prediction process. However, sometimes we have data that we want to fit specifically to a training set and then test on data set that we define. For example, training a model on years 2010-2015 and then testing a model on years 2016-2019.<\/p>\n

This tutorial walks through creating your own bespoke train\/test sets, fitting a model, and then making predictions, while circumventing the issues that may arise from not having the initial_split()<\/strong> object.<\/p>\n

Load the AirQuality Data<\/strong><\/p>\n

\r\nlibrary(tidyverse)\r\nlibrary(tidymodels)\r\nlibrary(datasets)\r\n\r\ndata("airquality")\r\n\r\nairquality %>% count(Month)\r\n<\/pre>\n
 <\/p>\n
Train\/Test Split<\/strong><\/p>\n
We want to use `tidymodels` to build a model on months 5-7 and test the model on months 8 and 9?<\/p>\n
Currently the initial_split()<\/strong> function only takes a random sample of the data.<\/p>\n
\r\nset.seed(192)\r\nsplit_rand <- initial_split(airquality ,prop = 3\/4)\r\nsplit_rand\r\n\r\ntrain_rand <- training(split_rand)\r\ntest_rand <- testing(split_rand) train_rand %>%\r\n  count(Month)\r\n\r\ntest_rand %>%\r\n  count(Month)\r\n<\/pre>\n
\"\"<\/a><\/p>\n
The strat<\/strong> argument within initial_split()<\/strong> only ensures that we get an even sample across our strat<\/strong> (in this case, Month).<\/p>\n
\r\nsplit_strata <- initial_split(airquality ,prop = 3\/4, strata = Month)\r\nsplit_strata\r\n\r\ntrain_strata <- training(split_strata)\r\ntest_strata <- testing(split_strata) train_strata %>%\r\n  count(Month)\r\n\r\ntest_strata %>%\r\n  count(Month)\r\n<\/pre>\n
\"\"<\/a><\/p>\n