Skip to content

I am working on a YouTube analytics project focused on the SNAKEHIPSUK channel. The project involves utilizing machine learning and data science techniques to predict columns in a CSV file containing relevant data.

Notifications You must be signed in to change notification settings

Shab00/SNAKEHIPSUK

Repository files navigation

# YouTube Analytics Project: SNAKEHIPSUK Channel

## snakeHipsAnalytics.ipynb
## snakeHipsAnalyticsPLayers.ipynb
This Jupyter Notebook contains a comprehensive analysis of Flagrant2 videos, including data preprocessing, feature selection, normalization, and various visualizations to gain insights into the dataset.

### Table of Contents
1. Introduction
2. Dataset
3. Installation
4. Usage
5. Analysis Steps
   - Selecting Relevant Columns
   - Normalization/Standardization
   - Correlation Matrix and Heatmap
   - Top 5 Videos by View Count
   - Total View Count by Day of the Week
   - Time Series of Video Views
   - Distribution of Video Durations
   - Word Cloud Visualization
6. Contributing
7. License

### Understanding the Data Columns
- title
- publishedAt_timestamp
- viewCount, likeCount, commentCount
- tagCount
- duration_seconds
- caption_False, caption_True
- definition_hd, definition_sd
- publishDayName

### Feature Engineering
This section focuses on creating new features from the existing ones to potentially improve the model's performance. Two new features are engineered.

### Creating Target Variable and Data Cleaning
This section defines the target variable for the machine learning model and performs some data cleaning.

### Defining the Model Architecture
This code defines a function `get_model` that creates and compiles a neural network model and prints a detailed classification report.

## requirements.txt
Contains the list of required packages for running the Jupyter Notebook.

# Key Model Performance Metrics

## Precision
- For class 0 (not popular): 0.92
- For class 1 (popular): 0.85

## Recall
- For class 0: 0.86
- For class 1: 0.92

## F1-Score
- Both classes: around 0.88

## Support
- Class 0: 14 samples
- Class 1: 12 samples

## Accuracy
- 0.88

The model demonstrates strong performance overall, with high accuracy and good precision and recall for both classes. The slight difference in precision and recall between the classes suggests that the model might be slightly better at identifying "popular" videos than "not popular" videos.

This README.md provides an overview of the YouTube analytics project for the SNAKEHIPSUK channel, detailing the content and purpose of each file in the repository.

About

I am working on a YouTube analytics project focused on the SNAKEHIPSUK channel. The project involves utilizing machine learning and data science techniques to predict columns in a CSV file containing relevant data.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published