This is a Text Data Analysis Project Involving (YouTube Case Study).

There are many Sentiment Packages such as Vader, Pacy. In this project i am using TextBlob which is a NLP library.

When considering Texblob Sentiment Analysis, there are two keys involved which are Polarity and Subjectivity.

-- Polarity: Which ranges from [-1 to +1] for negative and positive sentiments.

-- Subjectivity: When there are no Sentiments in a sentence.

Use try,except to handle error in your code.

The second package I'll be using to perform visualization on the sentiment Analysis is Wordcloud.

-- wordcloud:analysis give regard to the keyword with the bigger Font, therfore any keyword with this attribute has the higher priority.

-- wordcloud : data must also be stored in String nature before being passed.

wordcloud can be tricky when installing . irrespective of command prompt or conda prompt.. open and run as administrator and excute the following line of codes below.

-- for conda. conda install -c https://conda.anaconda.org/conda-forge wordcloud

-- For command prompt.

git clone https://github.com/amueller/word_cloud.git
cd word_cloud
pip install .

This are words that donot make any sense in Analysis. such as He, Him, Is, The.

-- wordcloud has parameters that removes this stopwords . ie (stopwords = reset(STOPWORDS).

-- !pip install emoji on Jupyter notebook.

-- pip install emoji on conda prompt or Cmd prompt open and run as administrator

After iterating the emoji_list, you need to compute it into frequncies, which means you need to come up with data in the form of Dict. this can be implemented completely from Scratch, or using count. But in this analysis we shall be using Collections models.

-- The first way to do this by using the OS, interating with the OS and using a path to access all the files.

-- The second way is Glob, which is consider to be the best way.

-- latin, UTF-8, iso-8859-1 (essential in reading complex data example is japanese data), cp-1252

The category file was clean and coverted to dict, then,the Category_name in the dict was map with category_id in the full_df and store in a column created in the full_df.

-- finding the maximum likes in the category can be archived with groupby as well but we used boxplot to visualized.

The three most important keys to consider when solving this kind of problem are Like_rate, Dislike-rate, Comment_rate.

In this case you can use Scatter plot or Regplot to check and also correlation and visualized it using heatmap

when using seaborn regplot in jupyter notebook always set ci=None, for it to execute.

In this project the data frame had only channel_tile and video_id, so we used a groupby function to group the video_id's according to there channel_title. and renamed the video_id's column to total_video according to there channel_title respectfully, the channel with the highest count of video_id had the trending vidoes.

We have to extracts all the punctuations marks from title, channel_title or tags, and this can be done completely from scratch or using python build in modules like Regular expression, Strings. in this project we shall use the string module.

--- To check if the punctuatons will affect the views,likes,dislikes or comments. use correlatoin to check the count punctuations from title, channel_title or tags on the views,likes,dislikes or comments.