Water Quality Analysis & Employee Data Tasks

Set of scripts for data processing, ML classification, and database integration for the club project.

1. Prerequisites

• Python 3.10+
• MySQL Server
• Virtualenv (recommended)

2. Installation

Clone the repo and install the requirements via pip.

we'll set up a virtual environment, and then install pandas in it. because many times installs conflict with global permissions on users' computers. Many times projects require different versions of dependencies, so thats why this is a requirement. but it should run fine on any system with pandas

python -m venv venv
source venv/bin/activate  # venv\Scripts\activate on Windows
pip install -r requirements.txt

TASK- 1:

Implements a Gaussian Naive Bayes model to predict water potability based on chemical features. The pipeline includes median-based null handling and exports results directly to a SQLite database.

TASK - 2:

As per the task, I made a commented program that replaces all NaNs in the salary column with the department-wise median.

Why is Grouped Imputation better than Global Imputation? IN a company people are paid differently in each department, if we replace all null values to the median of the entire slary column, we will underestimate the salary of a Project Mangager and overestimate the salary of a janitor, so thats why I took the median of the salary department wise, that way the null values will be replaced more accurately

TASK -3:

I first converted the timestamp from string to date-time so that i can sort by it as mentioned in the task, then i found and counted the number of duplicates in transaction id, stored it in a series, so that i can just give the length of series to tell how many duplicates there were, this wasnt in the task but it seems like valuable info. then i sorted the dataframe according to timestamp and removed all duplicates except the last.