• Description
• Methods
• Results

This project is an analysis of water quality tests using data from a Kaggle dataset.

The dataset provides 200 rows of water quality testing instances contemplating the following variables:

• pH
• Temperature (°C)
• Turbidity (NTU)
• Dissolved Oxygen (mg/L)
• Conductivity (µS/cm)

The aim of this project is to assess the data provided through an Exploratory Data Analysis to evaluate the physical and chemical parameters of the water in accordance with the Brazilian legislation regarding water potability (original portuguese text).

The eda.ipynb and testing.ipynb file are the notebooks for the EDA and for the testing of the module developed for the modelling respectively. The data/ folder includes the .csv and src/ contains the functions for preprocessing and modelling of a Random Forest Regressor.

The structure of the data can be viewed here:

A pairplot and a heatmap are useful in demonstrating the distribuition of data and the linear correlations betwen the variables:

The performance of the model is displayed in code:

[RandomForestRegressor(n_estimators=1000, random_state=42),
 {'r2': 0.7120092074666291,
  'mae': 0.26272562626261914,
  'rmse': 0.45081460320675126}]