Netflix Data Science Project: 6 Tasks on 8,790 Titles | Auspify Internship

I worked through the Netflix titles dataset (8,790 titles) for the Auspify Technologies Data Science Internship. The video covers all six tasks in under 90 seconds. What happens in each task: Data cleaning found 401 country values that were filler, three titles that Excel had turned into dates, and four duplicate rows. The EDA shows that 70% of the catalogue is movies, while the TV share of new titles grew from 25% to 34% between 2018 and 2021. A TF-IDF and cosine similarity recommender matches the right audience 84% of the time, up from 50% with genres alone. Eight forecasting models were compared, and ARIMA predicts about 180 new titles a month. An XGBoost classifier predicts a title's maturity level with 66.5% accuracy against a 46% baseline. Everything ends in an interactive Plotly dashboard with eight business recommendations. I used AI tools for help with parts of the code and the video edit. Chapters 0:00 Intro 0:09 Task 1: Data cleaning 0:19 Task 2: Exploratory data analysis 0:28 Task 3: Recommendation system 0:39 Task 4: Trend forecasting 0:51 Task 5: Classification 1:02 Task 6: Business dashboard Tools: Python, pandas, scikit-learn, XGBoost, statsmodels, Plotly #DataScience #MachineLearning #Python #Auspify #AuspifyInternship