Problem statement:
Data set of pumps installed in various places in the country to serve the water for their countrymen. It is a tedious and expensive task to maintain these pumps. This data set contains information such as the kind of pump, when it was installed, and how it is managed. Can you predict which pumps require repairs and which are not functional by using the given data set? A smart understanding of water point failure can improve maintenance operations and ensure that clean and safe water is available to these communities. The repository contains codes for training supervised learning models and some ensemble machine learning models to classify and predict functional and non-functional pumps in a country based on data available.
Topic - Failure Prediction using classification techniques
Objective: To Predict Failure of Water Pumps in Tanzania using standard classification techniques.
• Updated the data in order to remove null values using various imputing techniques and introducing suitable new labels
• Implemented Bagging classifier, Gradient Boosting classifier, Support Vector Classification and Decision Tree
• Assessed the performance of the models using accuracy as the performance metric to obtain the most suitable classifier