Therefore, I encoded or factorised categorical features. Import Data and Packages library(tidyverse) # data manipulation library(caret) # predictive modelling library(ot) # decision tree visualisation mics <- read_dta("MICS.dta")Īll categorical variables are text strings or characters, which machine learning models cannot really work with. Ever had a child or children who later died, child_died: Binary (Yes/No).Ever given birth, given_birth: Binary (Yes/No).Country, country: Categorical (Thailand/Mongolia/Laos).
Residence, urban: Categorical (Urban/Rural). Marital status, mstat: Categorical (Never/Former/Current). Country-specific wealth percentile, wealth: Numerical. Highest education attainment, edu: Categorical (Less than primary/Primary/Lower secondary/Upper secondary/Higher education). Currently using contraception, use: Binary (Yes/No). As a result, the dataset contains 58,356 cases and 9 variables: After a quick registration, I downloaded the 6th MICS datasets for three countries and performed data cleaning and merging. Because they are microdata for individual response data in surveys and censuses, users must register to access the database. The data are from the Multiple Indicator Cluster Surveys (MICS) published by the United Nations Children’s Fund (UNICEF). The project's goal is to build classification models to predict the use of contraception in Thailand, Mongolia, and Laos. Here I focus on the caret package (short for Classification And REgression Training) created by Max Kuhn. Gladly, there are several libraries available that attempt to streamline the process for building predictive models. #CARET IN R HOW TO#
I learnt how to programme in R and use various statistical packages, such as glm and randomForest, but it felt inefficient, having so many different packages. “ An Introduction to Statistical Learning: With Applications in R” or ISLR was my first book on predictive analytics, and I strongly recommend everyone interesting in machine learning to read the book.
0 kommentar(er)
