Time Series Forecasting on CoronaVirus Dataset
summary
Live work uses the coronavirus confirmed-cases time series and compares naive forecasting, moving average, linear regression, SVM, LSTM, ARIMA and Facebook Prophet with RMSEs reported: moving average ~42,000, linear regression ~35,000, SVM ~22,000, LSTM ~10,000, Prophet ~9,000, ARIMA ~4,000, so ARIMA best (RMSE ≈ 4,000). Code and dataset pointers are on the presenter’s GitHub; the talk shows plotted forecasts, training/validation splits and which approaches actually track short-term trends versus overfit: data: confirmed cases time series models: naive MA LR SVM LSTM ARIMA Prophet.
AI-written summary · about
chapters (28)
- 0:00Introduction
- 0:30Agenda
- 2:25History of pandemic
- 4:08Metrics
- 4:38Prediction and Forecasting
- 7:05Moving Average
- 7:28Window Size
- 7:49Graph
- 8:25Linear Regression
- 9:03Linear Regression Training
- 9:35Linear Regression Library
- 10:02Linear Regression Fit
- 10:23Graphing
- 12:23Point in 3D
- 12:38Classification
- 12:58Vector Machine
- 14:52Plot
- 15:16Why RNN
- 16:11RNN Architecture
- 16:27Vanishing Gradient
- 17:52LTSM
- 19:18Activation Function
- 20:06Forecasting
- 21:23Profit
- 21:51Plotting
- 22:06Results
- 22:29Questions
- 25:12Learnings
related talks
25mTimeSeries Forecasting with ML Algorithms and there comparisons
01h 26mApplying Statistical Modeling & Machine Learning to Perform Time-Series Forecasting
27mBeyond a Machine Learning Model
41mNeural Prophet – A powerful AI framework for Time Series Models
39mForecasting With Classical and Machine Learning Methods
37mAI/ML Strategies & Techniques for a Post Pandemic World
29mTime Series Forecasting Techniques in Python
22mCOVID-19: Data Science & Expertise