Predicting Hotel Cancellations with Machine Learning
summary
Uses a hotel bookings dataset with feature selection via ExtraTrees, heavy pandas manipulation and notebooks on the mg course and stats GitHub, then compares logistic regression, SVM and a neural net (500 epochs) to classify cancellations, giving AUC 74% (SVM), ~76% (NN). Time series work contrasts ARIMA on trending data with LSTM on jumpy series; LSTM predicts direction 80% of the time using five lagged weeks and 20 epochs. Amazon SageMaker is used for cloud runs.
AI-written summary · about
chapters (11)
- 0:00Start
- 9:05Accuracy
- 9:38Training Accuracy
- 13:40Time Series Analysis
- 15:39Constructing the Lcm Model
- 16:36Lcm Results
- 18:34No Free Lunch Theorem
- 19:38Data Architecture
- 20:29Easier To Coordinate Python Versions across Users
- 21:27Sample Workflow on Amazon Sage Maker
- 22:46Findings
related talks
34mMachine Learning to know if you R coming
44mMachine learning @ booking.com
15mMachine Learning with Python Case Study
40mMachine Learning
34mForecasting Airline Passengers Using Designer Machine Learning
25mTimeSeries Forecasting with ML Algorithms and there comparisons
30mDriving a Higher Level of Customer Support with Machine Learning
39mForecasting With Classical and Machine Learning Methods