TikTokClaimDetective: Claim or Opinion using ML

The aim of the project is to use machine learning to extract claims or propositions within TikTok videos and comments and classify each data point as a claim or opinion. To approach this task i will be using the PACE frame-work that is an accronym for a 4 stage approach Plan, Analyze, Construct and Execute.

Plan :

1- Goal: To determine whether a video contains a claim or whether it offers an opinion.

2- Strategies utilized :

Data Source : Gathered with in from the organization (first party) and provided as a csv.

KPIs : Would depend on the insights covered via statistical inferences and machine learning algorithms and their performnace.

Tools : Python, statsmodel, scipy, scikit-learn, Tableau and various others on need basis.

Deliverables for Plan stage : Dataset scrubbed for EDA, visualizations, statistical model, regression analysis, and/or machine learning.

3 - Deliverables :

Findings:

After looking at the dataset, the two variables that are most likely to help build a predictive model are claim_status and opinion_status.     We can create a classification model to differentiate the two variables, as instructed from the employer.

The mean is larger than the median, so the data is right-skewed. Therefore, it is recommended to consider applying a variable         
transformation, such as taking logarithms, prior to modeling.

Python file : Stage 1 Deliverable : Notebook

Analyze :

1- In this stage we will be covering the following processes:

  ETL (Extract, Transform, Load) as/if required.

  EDA (Exploratory Data Analysis) to uncover as much insights as possible.

  This stage also includes working with the stakeholders to create a subset of the insights discovered in the previous step. This subset         will    only include those insights that are to be pursued further.

2- Findings and Deliverables :

  Executive Summary : Stage 2 Deliverabe: Summary File

  Python File : Stage 2 Deliverable : Notebook

Construct :

1- This stage includes the following processes :

  Deciding what insights to pursue.

  Statistical Analysis

  Regression Analysis

  Utilizing Machine Learning Algorithm.

2- Starting with Statistical Analysis our next step is to find out if there is any statistical difference betweem the mean of video views count of verified and un-verified accounds. We do this by finding the p-value and if this less than our threshold of 5% we will reject the null hypothesis.

  Eexecutive Summary : Stage 3 Deliverables : Statistics Summary File

  Python File : Stage 3 Deliverables : Stats-Notebook

3- Earlier, we observed that if a user is verified, they are much more likely to post opinions. Since the end goal is to classify claims and opinions, it’s important to build a model that shows how to predict the behavior of the account type (verified) that tend to post more opinions. So, in this part of the project, we built a logistic regression model that predicts verified_status. The following files include the summary and notebook of the process.

  Executive Summary : Stage 3 Deliverables : Regression Analysis Summary

  Python File : Stage 3 Deliverables : Regression-Notebook

4- Now the final part comes in where we prepare the product. The following files contain the summary of the process and the model itself:

  Executive Summary : Stage 3 Deliverables : Final Product Summary File

  Python File : Stage 3 Deliverables : Product-Notebook

Execute :

1- This part of the framework is where we present the product to the stakeholders.

2- Our best model gave an accuracy of over 99%.