{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/c402a1d1e00240c0a35c02ec338ebb67\" frameborder=\"0\" width=\"1818\" height=\"1363\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":1363,"width":1818,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":1363,"thumbnail_width":1818,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/c402a1d1e00240c0a35c02ec338ebb67-e5d47f5f6f865e63.gif","duration":162.448,"title":"Data Science Assessment Overview: Weather Analysis and Predictions 🌦️","description":"Hi, I’m Purvansh Dave, a third-year ICT student, and this video is my submission for the PM Xerator data science assessment. I utilized the Global Weather Repository dataset from Kaggle, focusing on data cleaning, exploratory data analysis, and anomaly detection using Isolation Forest, which flagged about 5% of the data as anomalous. I explored various models, including ARIMA, Facebook Prophet, and XGBoost, with Facebook Prophet performing the best for seasonal predictions. I also conducted a unique analysis on feature importance, highlighting temperature, humidity, and wind speed as significant factors. You can find all the outputs and details in my GitHub repo, and I encourage you to check it out."}