KO
|
EN
gitlite โ search
Search
#python
#java
#python3
#arduino
#golang
#machine-learning
#rust
#html
#flask
#javascript
#seismology
#nodejs
DrivenData-PumpItUp
โ 12
Open GitHub โ
No description available.
Download README (.md)
Explore Similar Repositories
age-program
:
Complete Python Web Developer Course: First program (age in seconds)
D2D_cellular_network_simulator
:
Simulator used in the study "D2D frequency allocation using GADIA in a cellular network" (Matlab)
KMeanCoreset
:
KMean Coreset evaluation and computation.
parse-form
:
๐ Utility to easily parse forms in the browser.
color-hash.swift
:
Generates UIColor and NSColor from given string.
// repository documentation
Was this content helpful?
โ 0
(0 ratings)
Select Rating:
โ
โ
โ
โ
โ
Submit Feedback
Recent Feedback
×
Download README
Do you want to download the
README.md
file for
DrivenData-PumpItUp
?
Download (.md)
# DrivenData-PumpItUp This repository contains R code for the [Pump it Up: Data Mining the Water Table](http://www.drivendata.org/competitions/7/) competition on [Driven Data](http://www.drivendata.org). The data is provided by [Taarifa](http://taarifa.org) and the [Tanzanian Ministry of Water](http://maji.go.tz). The goal is to predict whether a water pump is functional, functional but needs repairs or non functional. I use [H2O](http://h2o.ai)'s random forest to get a score 0.821. I have uploaded my best (current) submission but not the data. Sign up at Driven Data to download the following files: * `SubmissionFormat.csv` * `Test set values.csv` * `Training set labels.csv` * `Training set values.csv` ## Read the data and do some preprocessing The first step is to read the data and set some values to missing (`NA` in R): See `read-data.md`. ## Engineer features The next step is to clean up the features (transform some, remove others) and possibly engineer some new features: See `transform-data.md`. ## Predict status with a random forest Use a random forest to predict the functionality status of pumps in the test set: See `predict-data.md`. ### Update Added a Makefile, which `spin`s the `R` scripts to produces the `md` files. See how to [Build a report based on an R script](http://yihui.name/knitr/demo/stitch/).
