buchacher/pumps

A multi-layer perceptron classifier to predict water pump repair status in Tanzania implemented using sklearn # Data Mining Tanzanian Water Pumps The aim was to construct and use artificial neural networks to solve a water-pump classification problem with real-world data from Tanzania. The solution, which predicts the repair status of water pumps, relies on a multi-layer perceptron classifier and entails a logistic regression classifier as a baseline for evaluation purposes. Both were implemented using sklearn. Different methods to deal with imbalanced data were implemented and evaluated, among them SMOTE-NC using imbalanced-learn. ### Compiling and Running Instructions **Begin by copying the datasets into the `place-data-here` directory**. Navigate into the ```pumps``` directory: ```shell script cd pumps ``` Set up a virtual environment within the directory: ```shell script python3 -m venv my_env ``` Activate the virtual environment: ```shell script source my_env/bin/activate ``` Install the requirements to your virtual environment via pip: ```shell script pip install -r requirements.txt ``` To run via ```main.py```, your command should take the following form: ```shell script python3 main.py ``` For example, pass... ```shell script python3 main.py task1 test ``` ...to predict on the task 1 test set. The output should be: ```shell script Test set classification accuracy: 0.634 ``` ### Valid Standard Commands ```shell script python3 main.py task1 train python3 main.py task1 test python3 main.py task1 predict python3 main.py task2 train python3 main.py task2 test python3 main.py task2 predict python3 main.py task3 train python3 main.py task3 test python3 main.py task3 predict python3 main.py task4 train python3 main.py task4 test python3 main.py task4 predict ``` ### Self-Contained Scripts In addition, a number of self-contained scripts can be executed as follows: ```shell script python3 baseline_task1.py python3 baseline_task2.py python3 baseline_task3.py python3 baseline_task4.py python3 task1_explore.py python3 task2_explore.py python3 task3_explore.py p …