Logo Lanfrica

kykar/water_pump_condition

Domain:

environment and energy

Record type:

software
Creator:
kyk
Host:
Predict functionality of water pumps in Tanzania # Predicting Waterpoint Condition The purpose of this project is to predict the functionality of drinking water wells in Tanzania. The data comes from drivendata.org ## Directory Tree: │--water_pump_condition.Rproj │ │--data │------test.csv │------test_raw.csv │------train.csv │------train_labels.csv │------train_raw.csv │ │--models │------feature_elimination.rds │------knn.rds │------multinom.rds │------rf_mtry.rds │------xgb_tune.rds │ │--reports_and_figures │------wpc_report.docx │------wpc_report.html │------wpc_report.Rmd │ │--scripts │------cor2Fun.R │------wpc_eda.R │------wpc_prelim_models.R ### data - test.csv preprocessed test data - test_raw.csv raw test data - train.csv preprocessed train data - train_labels.csv labels for target for train data - train_raw.csv raw train data ### models models are created in /scripts/wpc_prelim_models.R - feature_elimination.rds recursive feature elimination with random forest - knn.rds k-nearest neighbors - multinom.rds multinomial logistic regression - rf_mtry.rds random forest tuning mtry (number of variables randomly sampled at each split) - xgb_tune.rds XGBoost tuning various parameters ### reports_and_figures - report in .Rmd, .html, and .docx formats - images used in report ### scripts - cor2Fun.R function that computes the correlation coefficients between different data types - wpc_eda.R exploratry data analysis and data preprocessing - wpc_prelim_models.R preliminary modeling