james-wachuka / speeddating_r Goto Github PK

A classification problem in R for determining the possibility of a second date based on the outcome of the first date.A data set for this task is provided:speeddating.csv

The target variable second_date was created for use in this task.Amelia function was used to impute the missing values.Non-numeric values have been encoded for ease of use in analysis and modelling.

decision trees, box plots, barcharts, jitter plots were used to analyse data, with the aim of discovering the underlying information.A detailed analysis is provided in the R notebook.

Random forest was preferred for feature selection as compared to correlation matrix because it gives a simpler result that is easy to study.Logistic regression,KNN and Bayesian Classifier were used for this task with random sampling of rows.The models were evaluated based on AUC score.

This R notebook contains the detailed presentation of the whole task.