Water Quality Prediction

Introduction About the Data :

• The dataset includes information about various substances present in water, typically measured in units of concentration per liter.

All attributes are numeric variables and they are listed below :

• A description of the data attributes.
• aluminium - dangerous if greater than 2.8
• ammonia - dangerous if greater than 32.5
• arsenic - dangerous if greater than 0.01
• barium - dangerous if greater than 2
• cadmium - dangerous if greater than 0.005
• chloramine - dangerous if greater than 4
• chromium - dangerous if greater than 0.1
• copper - dangerous if greater than 1.3
• flouride - dangerous if greater than 1.5
• bacteria - dangerous if greater than 0
• viruses - dangerous if greater than 0
• lead - dangerous if greater than 0.015
• nitrates - dangerous if greater than 10
• nitrites - dangerous if greater than 1
• mercury - dangerous if greater than 0.002
• perchlorate - dangerous if greater than 56
• radium - dangerous if greater than 5
• selenium - dangerous if greater than 0.5
• silver - dangerous if greater than 0.1
• uranium - dangerous if greater than 0.3
• is_safe - class attribute {0 - not safe, 1 - safe}

Target variable: Here is_safe is the Dependent variable.

Dataset Source Link : https://www.kaggle.com/datasets/mssmartypants/water-quality

Goal of this project:

The objective is to categorize the provided instances into one of two distinct categories and predict the percentage indicating the quality of water being good.

Approach for the project:

1. Data Preprocessing:
   • In this initial stage, we identify the null values. The #NUM! values are replaced with NaN, and then the NaN values are dropped, as there are very few of them.
2. Data Transformation:
   • In this stage, standard scaling is performed on the complete dataset except for the target variable.
3. Model Training:
   • In this phase, the model was trained using a Random Forest Classifier, which achieved an accuracy of 95.19%.
   • In a binary classification problem, the function predict_proba was utilized to compute the probabilities for the given x_test data.
4. Flask App Creation:
   • The Flask library is used to develop a web application that serves as a user interface for predicting water quality as a percentage.

Postman Testing of API :

User Interface:

Predicted Output In Percentage: