1002_Project Presentation - https://drive.google.com/file/d/1fZ9JZ01cs31K3R2xb7co0nrWrmB90r9m/view?usp=sharing

README: Data Science Project Phases

This README outlines the various phases of the project, including Machine Learning, Data Cleaning, Exploratory Data Analysis (EDA), SQL Integration, and Power BI Visualizations. Each phase is critical for building an end-to-end data-driven solution.

---

1. Machine Learning Phase (ML_Phase)

Purpose:

To build and optimize a predictive model to achieve actionable insights from the data.

Key Steps:

• Feature Engineering:
  • Transformed raw data into meaningful features (e.g., creating age groups and binning balance).
• Model Selection:
  • Used the Random Forest Classifier due to its robustness and ability to handle imbalanced data.
• Hyperparameter Tuning:
  • Utilized GridSearchCV to optimize parameters like the number of trees and max depth, with a focus on maximizing the F1 score.
• Evaluation Metrics:
  • Prioritized F1 score (achieved 67.75) to balance precision and recall, considering the dataset's imbalanced nature.

Output:

A tuned machine learning model ready for deployment.

---

2. Data Cleaning Phase (Cleaning_Phase)

Purpose:

To ensure the dataset is free of inconsistencies, outliers, and missing values to improve downstream analysis and modeling.

Key Steps:

• Outlier Detection:
  • Identified and removed extreme values using statistical methods.
• Missing Data Handling:
  • Imputed missing values (e.g., using mean or median for numerical features and mode for categorical features).
• Tools Used:
  • pandas, numpy

Output:

A clean and reliable dataset ready for analysis.

---

3. Exploratory Data Analysis Phase (EDA_Phase)

Purpose:

To uncover patterns, trends, and relationships in the data to inform decision-making.

Key Steps:

• Visualizations:
  • Used seaborn and matplotlib to create distribution plots, heatmaps, and boxplots.
• Correlation Analysis:
  • Examined feature relationships to guide feature selection.
• Data Issues:
  • Identified and resolved missing values and duplicates.

Output:

Insights and visualizations that guided feature engineering and model design.

---

4. SQL Phase

Purpose:

To integrate structured querying for database management and advanced data manipulation.

Key Steps:

• Data Extraction:
  • Retrieved relevant data using SQL queries.
• Data Aggregation:
  • Used aggregation functions (e.g., GROUP BY, SUM, AVG) for summary statistics.
• Filters:
  • Applied WHERE and HAVING clauses to filter and refine data subsets.
• Database Creation:
  • Created a new database and table to store training data:

Output:

Well-structured data outputs for downstream analysis.

---

5. Power BI Visualization Phase

Purpose:

To create dynamic dashboards and visualizations for presenting actionable insights.

Key Steps:

• Data Import:
  • Imported cleaned and preprocessed data into Power BI.
• Dashboard Design:
  • Created interactive dashboards with key performance indicators (KPIs).
  • Examples: Sales trends, customer segmentation, and correlation heatmaps.
• Interactivity:
  • Added slicers and filters to allow users to explore data dynamically.

Output:

An interactive Power BI dashboard providing a user-friendly interface for data exploration.

---

General Notes:

• The project workflow is modular, enabling iterative improvements.
• Tools Used: Python, pandas, seaborn, sklearn, SQL, Power BI
• Key Focus: Delivering insights that are actionable and aligned with business goals.